The Economist - Edição 9022 - (07 Janeiro 2017)
The Economist - Edição 9022 - (07 Janeiro 2017)
Now we’re
talking
Voice computing
comes of age
The Economist January 7th 2017 5
Contents
7 The world this week Asia
28 Ageing in Japan
Leaders Cities vie for the young
9 Voice technology 29 Japan’s elderly workers
Now we’re talking Keep on toiling
10 Japan’s economy 29 Alcohol in Indonesia
The second divine wind Calls for a ban
10 Trumponomics 30 New Zealand’s national
Men of steel, houses of parks
cards Lord of the ker-chings
31 Banyan Theresa Maybe It is still
11 Fixing failed states
Selling Malaysians down unclear what Britain’s new
First peace, then law
the river prime minister stands
12 British politics for—perhaps even to her:
On the cover Theresa Maybe leader, page 12. The making
Voice technology is making China and meaning of a prime
computers less daunting and Letters 32 Selection year minister, pages 18-20. The
more accessible: leader, page A reshuffle looms sudden departure of Britain’s
13 On China, management,
9. Computers have got much 33 Literature man in Brussels lays bare the
elections, nuclear power,
better at translation, voice Meet China’s Shakespeare lack of Brexit plans, page 35.
Japan, the elderly,
recognition and speech The “WTO option” for Britain is
economists
synthesis. But they still far from straightforward: Free
don’t understand what Britain exchange, page 58. The first
language means: Technology Briefing 34 Crime crop of Brexit books includes
Quarterly, after page 36 18 Theresa May How low can it go? entries rich in detail and
Steering the course 35 Brexit preparations analysis, page 63
Rogers and out
The Economist online 35 Foreign aid
United States
Daily analysis and opinion to A stingy new year
supplement the print edition, plus
21 Inequality
Fat tails 36 Bagehot
audio and video, and a daily chart Pierogi and integration
Economist.com 22 Congressional ethics
E-mail: newsletters and Old bog, new tricks
mobile edition 23 Recruiting police officers Technology Quarterly:
Economist.com/email The force is weak Language
Print edition: available online by 23 Gun laws Finding a voice
7pm London time each Thursday Still standing After page 36
Economist.com/print 24 Charleston
Audio edition: available online Cobblestones and bones Middle East and Africa Turkey torn apart
to download each Friday 24 Markets for tickets 37 South Africa’s schools The murderous Islamic State
Economist.com/audioedition Battling bots Bottom of the class attack on a nightclub widens
the secular-religious divide,
25 Lexington 38 Astronomers v sheep page 42
Learning to love Trumpism farmers in South Africa
Stars and baas
The Americas 38 Zimbabwe’s sex trade
Volume 422 Number 9022
Less stigma, more
26 Brazil’s prisons competition
Horror in the jungle
Published since September 1843 39 Iraq’s long war
to take part in "a severe contest between 27 Bolivia A goody and Abadi
intelligence, which presses forward, and
an unworthy, timid ignorance obstructing
Run, Evo, run
our progress."
40 America and Israel
Unsettled by Trump
Editorial offices in London and also:
Atlanta, Beijing, Berlin, Brussels, Cairo, Chicago, 41 Israel’s divisions
Lima, Mexico City, Moscow, Mumbai, Nairobi,
New Delhi, New York, Paris, San Francisco,
Convicting a soldier
São Paulo, Seoul, Shanghai, Singapore, Tokyo,
Washington DC
Failed states How to save
nations from collapse: leader,
page 11. The lessons from
Afghanistan and South Sudan,
page 46. Why South Africa has
one of the world’s worst
education systems, page 37
encroach and consumers fret 54 Bank capital forecasts Latin America US $289 (plus tax)
© 2017 The Economist Newspaper Limited. All rights reserved. Neither this publication nor any part of it may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or
otherwise, without the prior permission of The Economist Newspaper Limited. The Economist (ISSN 0013-0613) is published every week, except for a year-end double issue, by The Economist Newspaper Limited, 750 3rd Avenue, 5th Floor New York, NY 10017.
The Economist is a registered trademark of The Economist Newspaper Limited. Periodicals postage paid at New York, NY and additional mailing offices. Postmaster: Send address changes to The Economist, P.O. Box 46978, St. Louis, MO 63146-6978, USA.
Canada Post publications mail (Canadian distribution) sales agreement no. 40012331. Return undeliverable Canadian addresses to The Economist, PO Box 7258 STN A, Toronto, ON M5W 1X9. GST R123236267. Printed by Quad/Graphics, Hartford, WI. 53027
The Economist January 7th 2017 7
The world this week
Barack Obama expelled 35 Amazonas left 56 inmates The European Central Bank raise tariffs, portending what
Russian diplomats and im- dead. Some were decapitated; raised its estimate of the capi- may be one of his biggest fights
posed new economic sanc- severed limbs were stacked by tal shortfall at Monte dei with Mr Trump.
tions in retaliation against the entrance. Paschi di Siena to €8.8bn
Russian hackers’ interference ($9.1bn). The troubled Italian Luis Videgaray was rehabilitat-
in America’s election. Ameri- Odebrecht, a Brazilian con- bank has requested a bail-out ed in Mexico’s government by
can intelligence agencies say struction company, and Bras- from the government after being appointed foreign min-
that Russia released stolen kem, a petrochemical firm in running out of time to raise ister. Mr Videgaray resigned as
e-mails of Democratic Party which it owns a stake, pleaded new capital privately. finance minister after suggest-
officers in order to aid the guilty to bribing officials and ing that Donald Trump visit
campaign of Donald Trump. political parties to win con- Shortly before Christmas, Mexico last year, a hugely
Vladimir Putin declined to tracts in Latin American and Deutsche Bank agreed to pay unpopular move at the time.
strike back, winning praise African countries. The compa- $7.2bn to settle with America’s
from Mr Trump. nies agreed to pay a penalty of Department of Justice for Japan’s prime minister, Shinzo
at least $3.5bn, the largest mis-selling subprime mortgage Abe, paid his first visit to the
A gunman attacked a night- settlement ever in a global securities, about half the American naval base at Pearl
club in Istanbul during New bribery case. amount the regulator had Harbour. He expressed “sin-
Year’s Day festivities, killing at initially sought. Credit Suisse cere and everlasting condo-
least 39 people. Islamic State Stockmarkets had a good agreed to pay $5.2bn to resolve lences” to those who died in
claimed responsibility. Turk- 2016. The S&P 500 rose by 10% claims. But Barclays rejected a Japan’s attack on it 75 years
ish religious authorities who over the 12 months and the settlement, prompting the ago. Soon after, however, his
had criticised new year’s Dow Jones by 13%. The FTSE department to file a lawsuit. defence minister, Tomomi
celebrations as un-Islamic 100 recovered from its Brexit Inada, paid a visit to Yasukuni
condemned the attack. It came wobbles to end 14% up; Rus- Ford made a U-turn when it Shrine in Tokyo where Japa-
two weeks after a policeman sia’s RTS index soared after the scrapped plans for a new nese war criminals are hon-
shouting “Don’t forget Alep- election of Mr Trump to finish factory in Mexico to build oured among the war dead.
po!” fatally shot the Russian 52% higher; and Brazil’s Bo- compact cars, and diverted
ambassador to Turkey. vespa rose by 39%, despite, or some of the investment to a The British government ap-
because of, the defenestration plant near Detroit to produce pointed Sir Tim Barrow, a
Relations between Israel and of the president. But Italy’s electric vehicles. Ford stressed former ambassador to Russia,
America became strained main index fell by 10%, and that this was a commercial as its new ambassador to the
when John Kerry, the soon-to- China’s Shanghai Composite decision. Donald Trump had EU, three months before it is
retire secretary of state, said never fully recovered from its criticised the proposed Mex- due to trigger negotiations
that the Israeli government turbulent start to 2016, ending ican factory when he cam- over Brexit. This followed the
was undermining the pros- the year12% lower. paigned on the theme of sav- early exit of Sir Ivan Rogers
pects for a “two-state solution” ing American jobs. from the job. His resignation
with the Palestinians. His Donald Trump picked Jay note decried “muddled think-
comments came soon after Clayton, a legal expert on Meanwhile, Paul Ryan, the ing” by ministers.
America abstained in the UN mergers and acquisitions, to be most senior Republican in the
Security Council vote that the next head of the Securities House of Representatives, said Other economic data and news
criticised Israel’s construction and Exchange Commission. that Congress was not going to can be found on pages 68-69
of settlements.
A NY sufficiently advanced
technology, noted Arthur C.
Clarke, a British science-fiction
Although deep learning means that machines can recog-
nise speech more reliably and talk in a less stilted manner, they
still don’t understand the meaning of language. That is the
writer, is indistinguishable from most difficult aspect of the problem and, if voice-driven com-
magic. The fast-emerging tech- puting is truly to flourish, one that must be overcome. Comput-
nology of voice computing ers must be able to understand context in order to maintain a
proves his point. Using it is just coherent conversation about something, rather than just re-
like casting a spell: say a few sponding to simple, one-off voice commands, as they mostly
words into the air, and a nearby device can grant your wish. do today (“Hey, Siri, set a timer for ten minutes”). Researchers
The Amazon Echo, a voice-driven cylindrical computer that in universities and at companies large and small are working
sits on a table top and answers to the name Alexa, can call up on this very problem, building “bots” that can hold more elab-
music tracks and radio stations, tell jokes, answer trivia ques- orate conversations about more complex tasks, from retrieving
tions and control smart appliances; even before Christmas it information to advising on mortgages to making travel ar-
was already resident in about 4% of American households. rangements. (Amazon is offering a $1m prize for a bot that can
Voice assistants are proliferating in smartphones, too: Apple’s converse “coherently and engagingly” for 20 minutes.)
Siri handles over 2bn commands a week, and 20% of Google
searches on Android-powered handsets in America are input When spells replace spelling
by voice. Dictating e-mails and text messages now works reli- Consumers and regulators also have a role to play in determin-
ably enough to be useful. Why type when you can talk? ing how voice computing develops. Even in its current, rela-
This is a huge shift. Simple though it may seem, voice has tively primitive form, the technology poses a dilemma: voice-
the power to transform computing, by providing a natural driven systems are most useful when they are personalised,
means of interaction. Windows, icons and menus, and then and are granted wide access to sources of data such as calen-
touchscreens, were welcomed as more intuitive ways to deal dars, e-mails and other sensitive information. That raises pri-
with computers than entering complex keyboard commands. vacy and security concerns.
But being able to talk to computers abolishes the need for the To further complicate matters, many voice-driven devices
abstraction of a “user interface” at all. Just as mobile phones are always listening, waiting to be activated. Some people are
were more than existing phones without wires, and cars were already concerned about the implications of internet-connect-
more than carriages without horses, so computers without ed microphones listening in every room and from every
screens and keyboards have the potential to be more useful, smartphone. Not all audio is sent to the cloud—devices wait for
powerful and ubiquitous than people can imagine today. a trigger phrase (“Alexa”, “OK, Google”, “Hey, Cortana”, or
Voice will not wholly replace other forms of input and out- “Hey, Siri”) before they start relaying the user’s voice to the
put. Sometimes it will remain more convenient to converse servers that actually handle the requests—but when it comes
with a machine by typing rather than talking (Amazon is said to storing audio, it is unclear who keeps what and when.
to be working on an Echo device with a built-in screen). But Police investigating a murder in Arkansas, which may have
voice is destined to account for a growing share of people’s in- been overheard by an Amazon Echo, have asked the company
teractions with the technology around them, from washing for access to any audio that might have been captured. Ama-
machines that tell you how much of the cycle they have left to zon has refused to co-operate, arguing (with the backing of pri-
virtual assistants in corporate call-centres. However, to reach vacy advocates) that the legal status of such requests is unclear.
its full potential, the technology requires further break- The situation is analogous to Apple’s refusal in 2016 to help FBI
throughs—and a resolution of the tricky questions it raises investigators unlock a terrorist’s iPhone; both cases highlight
around the trade-off between convenience and privacy. the need for rules that specify when and what intrusions into
personal privacy are justified in the interests of security.
Alexa, what is deep learning? Consumers will adopt voice computing even if such issues
Computer-dictation systems have been around for years. But remain unresolved. In many situations voice is far more conve-
they were unreliable and required lengthy training to learn a nient and natural than any other means of communication.
specific user’s voice. Computers’ new ability to recognise al- Uniquely, it can also be used while doing something else (driv-
most anyone’s speech dependably without training is the lat- ing, working out or walking down the street). It can extend the
est manifestation of the power of “deep learning”, an artificial- power of computing to people unable, for one reason or an-
intelligence technique in which a software system is trained other, to use screens and keyboards. And it could have a dra-
using millions of examples, usually culled from the internet. matic impact not just on computing, but on the use of language
Thanks to deep learning, machines now nearly equal humans itself. Computerised simultaneous translation could render
in transcription accuracy, computerised translation systems the need to speak a foreign language irrelevant for many peo-
are improving rapidly and text-to-speech systems are becom- ple; and in a world where machines can talk, minor languages
ing less robotic and more natural-sounding. Computers are, in may be more likely to survive. The arrival of the touchscreen
short, getting much better at handling natural language in all was the last big shift in the way humans interact with comput-
its forms (see Technology Quarterly). ers. The leap to speech matters more. 7
10 Leaders The Economist January 7th 2017
Japan’s economy
The strong dollar has given Abenomics another chance. Now corporate Japan must do its bit
Trumponomics
The president elect’s team needs to realise that America’s economy is not like a steel mill
2 trade, sees the decline of America’s steel industry as emblem- ducers (in 2016 the Obama administration placed a tariff of
atic of how unfair competition from China has hurt America. 522% on cold-rolled Chinese steel), as has the European Union.
But the steel business is not a model for trade policy in gen- Yet this way of thinking fails to deal with the question of
eral and companies are capable of being tricksy, too. Mr Trump whether an ample supply of cheap steel courtesy of a foreign
may simply be looking for good headlines, but if he wants government is really so terrible: it benefits American firms that
more, his plans threaten to be an expensive failure. consume steel—and they earn bigger profits and employ more
people as a result. Moreover, trade in most goods and services
The miller’s tale is not like steel. America’s biggest import from China is electri-
One reason is that Paul Ryan, the Speaker of the House of Rep- cal machinery. China’s government does not subsidise the
resentatives, said this week that Congress would not be raising overproduction of iPhones which are then dumped on the
tariffs. Executive orders are bad politics and can get Mr Trump market, causing iPhone-makers in America to be laid-off. In-
only so far. Another is that Ford’s plans are not as simple as stead, a smartphone might be designed and engineered in Cal-
they look. It will still build its new small car in Mexico—at an ifornia and assembled in China, using components made or
existing plant (see page 51). But above all, Mr Trump gravely un- designed in half a dozen Asian and European countries, using
derestimates the complexity of messing with tariffs. metals from Africa. Likewise, every dollar of Mexican exports
The men of steel are right to complain about China. Its gov- contains around 40 cents of American output embedded
ernment has indeed subsidised its steelmakers, leading to a within it. For producers of such goods, tariffs would be a costly
glut that was dumped on the world market. Successive Ameri- disaster. American steelmakers might seek out government
can governments have put up tariffs to protect domestic pro- protection. Apple and its kind will not. 7
British politics
Theresa Maybe
After six months, what Britain’s new prime minister stands for is still unclear—perhaps even to her
Job description:
The incumbent will lead and coordinate a team of economists analysing economic
developments in Luxembourg and the euro area, conducting research on topics
pertaining to central banking, including monetary policy, and analysing public
inances, with a particular focus on Luxembourg. His/her responsibilities will also
include representing the Bank in high level national and international meetings.
The incumbent will report directly to the Governor.
Main tasks and responsibilities:
• Provide advice to the Governor on monetary policy and to Management in
general in terms of economic analysis and research;
• Develop the department’s work programme, with a particular emphasis on the
strategic direction of its research activities;
• Organize, supervise and assess the department’s work, in particular its
contribution to the BCL’s economic publications and to the department’s
research output;
• Develop research partnerships with universities, research institutes, think
tanks and other central banks.
Your profile:
• PhD in economics or M.A. in economics with extensive experience in research
and economic analysis;
• Experience in conducting and supervising research related to central banking
with a strong publication record;
• Solid knowledge of the Eurosystem’s monetary policy framework;
• Excellent command of English. French or German will be considered as an
advantage;
• Ability to communicate with peers, managers and policymakers; strong sense
of efficiency, organization and time management.
To apply, please email your cv and motivation letter to app8@egonzehnder.com
before February 15, 2017.
The Opportunity
We are currently looking for a Director, Research and Evidence to join us in our London office.
This is a new senior Leadership role in which you will provide global leadership to ensure a step
change in the rigour, relevance, and strategic direction of our global evidence base, so that we
can fully leverage our $2bn international programming portfolio across 120 countries; ensure
our ambitions for children, and become a knowledge leader on what works for children.
In order to be successful you will have:
• Extensive experience in applied implementation of data and evaluation systems to
improve frontline practice in the development/humanitarian context.
• Extensive skills and experience in designing, managing and communicating quantitative
and mixed method research and evaluation, using rigorous designs
• Demonstrable experience in the areas of research and evidence strategy development
and implementation
• Proven leadership abilities, with the ability to mobilise and motivate individuals outside
of your reporting line in a highly matrixed and federated environment
On a personal level you will be an inclusive and confident leader, with the gravitas to inspire
and influence a wide range of stakeholders, and a belief in the mission and values of Save the
Children.
This role offers a competitive salary, in the context of the sector, and a company pension
scheme.
The organisation
It’s an exciting time at Save the Children as we start the implementation of our new 15-year
global strategy – Ambition for Children 2030 – which focuses on achieving three Breakthroughs:
no child dies from preventable causes before their fifth birthday, all children learn from a quality
basic education, and violence against children is no longer tolerated.
Save the Children is a federated Membership organisation with 29 Members who are based
across the globe and provide Save the Children International with the funds to carry out our
international programming activities.
Application information
Please apply using a cover letter and CV explaining why you would be suited to this role, and do
also include your current and expected salary. Please send this all as a single document. A copy
of the full role profile can be found at www.savethechildren.net/jobs
We need to keep children safe, so our selection process reflects our commitment to the
protection of children from abuse.
2 by the moral and practical duties imposed ers cite her experience of diabetes—the ing gay marriage she shared. But she ex-
by the life of the church; all were thereby prime minister must inject herself with in- cluded and ignored those—like Jeremy
furnished with an unflashy, serious and sulin several times a day. But the best ex- Browne and Norman Baker, Ms Feather-
cautious character. planation is her career as a woman educat- stone’s two successors in the department—
Her vicarage childhood lives on in Mrs ed at a provincial grammar-school (the with whom she did not.
May’s very English traits. She drinks Earl granddaughter of domestic servants, no She clashed with Michael Gove, then
Grey tea, reads Jane Austen, watches James less) in a party dominated by public- the education secretary, over measures to
Bond films, regularly attends church in her school boys given to cavalier confidence deal with extremism in schools and with
constituency (Maidenhead, a posh town in and clever-clever plans. When her allies Mr Osborne over immigration—she want-
the Thames valley) and adores cricket. Ech- praise Mrs May’s methodical style and her ed to tighten up Britain’s student visa re-
oes of this can be seen in her leadership. disdain for chummy, informal “sofa gov- gime. She was typically one of the last min-
Anglicanism often combines stormy, king- ernment”, they are channelling her long- isters to agree on her department’s budget
dom-of-God language with a restrained held exasperation with the know-it-all in the annual financial round. She also had
conservative culture: hymns about crusad- posh boys—particularly Mr Cameron and a run-in with Boris Johnson, then mayor of
ers and the devil belted out before tea and George Osborne, his chancellor. London, over three water cannon he
biscuits. In her first months as prime minis- The prime minister has little time for bought without seeking the Home Office’s
ter Mrs May, too, has been bolder in her the parliamentary village, avoiding its bars necessary—and, in the event, withheld—
rhetoric than in her actions—big ideas have and tea rooms, declining dinner-party invi- approval. The incident serves her inner cir-
received little follow-through, or been tations in London—let alone in Brussels, or cle as a house parable showing the perfidy
dropped altogether. There is a touch of her Washington, DC. She is the opposite of cos- of civil servants (who talked Mr Johnson
cricketing hero, Geoffrey Boycott, about mopolitan. “If you believe you’re a citizen into the idea), the folly of ill-scrutinised de-
her too. It is hard not to detect her admira- of the world, you’re a citizen of nowhere,” cisions, the danger of informal structures
tion for the stolid style of the Yorkshire she told her party conference in October. and the comeuppance of those who do not
batsman in her matter-of-fact demeanour. She struggles with the small talk that oils do things Mrs May’s way.
When her aides say “She just gets on with diplomatic (and cabinet) wheels. The Euro- In Downing Street Mrs May has im-
the job” it is the sort of praise their boss pean Council summit on December 16th posed the centralised, formal working
would like. saw the prime minister fiddling awkward- practices that she honed at the Home Of-
A social reformism rooted in her Angli- ly with her cuffs as fellow leaders air- fice. The day is governed by the 8.30am
can upbringing and practice (“part of who I kissed behind her. She is far more at home meeting, a shoeless free-for-all under Mr
am and therefore how I approach things”, in her constituency on the banks of the Cameron that now has a strict invitation
she has said) has been a constant of her ca- Thames. Her house in the village of Son- list. Blue-sky thinking and speculation
reer. When the voters of Maidenhead first ning sits by what Jerome K. Jerome, a Vic- about the headlines that evening are out;
sent her to Westminster in 1997 she was, in torian humorist, described as “the most firm instructions to staffers are in. In the
this respect, to the left of her party. In 2002 fairy-like little nook on the whole river”. prime minister’s office a table and chairs
she warned her colleagues and their sup- Here, in her natural habitat, she is by all ac- (and vases of hydrangeas) have replaced
porters that they had become known as counts witty, relaxed and gregarious. the sofa. Ministers and staffers must sub-
“the nasty party”. The following year, as mit papers earlier than under Mr Cameron,
shadow transport minister, she argued for Ordering their estate to allow her to work through them late in
more state intervention in the economy, a Mrs May’s time running the Home Office, a the evening (he would do them the next
more nuanced relationship with trade un- department institutionally obsessed with day). The whole machine is run by a small,
ions and limits on fat-cat excesses. order and control, earned her a reputation powerful team centred on her two chiefs-
All of this lives on in her premiership. for inscrutability, formality and obsession of-staff, Fiona Hill and Nick Timothy.
When, having lost the Brexit referendum, with detail (“she was always asking for Cabinet and sub-cabinet meetings are
Mr Cameron resigned, Mrs May enumerat- more papers in her red box,” says one lieu- venues for serious discussion, not Potem-
ed the inequities of modern Britain as she tenant). She worked well with people with kin forums with pre-decided outcomes.
launched her campaign to succeed him: whom she had things in common, like Having for the most part distributed minis-
boys born poor die nine years earlier than Lynne Featherstone, the Liberal Democrat terial portfolios evenly between Leavers
others; children educated in state schools minister whose commitment to introduc- and Remainers, Mrs May appointed three 1
are less likely to reach the top professions
than those educated privately; many
women earn less than men. The one-party state
Conservative party conference. Article 50 European
When she became prime minister she Britain, 2016 Pledge to trigger Article 50 Supreme Council
by April 2017 Court case Summit
repeated some of these “burning injus- Brexit vote Theresa May becomes prime minister
tices” on the steps of Downing Street. She 60
Who would do better as prime minister? Mrs May backs new
has talked up a new generation ofstate-run % replying: Theresa May runway at Heathrow
grammar schools (schools, like the one she 50
VOTING INTENTION
attended, that are allowed to select their % replying: CONSERVATIVE
pupils through competitive exams) to give 40
clever children from poor backgrounds a
LABOUR
leg up. She has hinted at worker represen- 30
tation on company boards; she has lament-
ed the effect of the Bank of England’s low
20
interest rates on savers. Jeremy Corbyn
21 Labour
Mrs May patently stands apart from front High Court rules
benchers Parliament Mrs May retreats 10
many of her colleagues in ways that go be- resign Labour leadership contest should vote from workeers--on-
Tory leadership on Article 50 boards pledgee
yond this reformism; there is a social dis- contest
tance, too. Some say it has to do with the 0
June July August September October November December
isolating shock of losing both of her par-
ents when she was relatively young. Oth- Sources: ICM; Ipsos MORI; Opinium; YouGov
20 Briefing Theresa May The Economist January 7th 2017
2 people who, unlike her, campaigned for plunged into a legal bunfight to prevent it. gave.” Mr Boycott, one feels, might approve
Brexit to the departments most concerned As the Deloitte memo put it, she seems to such dogged defensiveness; but few would
with bringing it about—Mr Johnson to the have no coherent plan for Brexit, her gov- look to him for lessons on team building.
Foreign Office, Liam Fox to a new Depart- ernment is “struggling” and still she is On coming to power it was not enough
ment for International Trade and David prone to “drawing in decisions and details for Mrs May to fire Mr Osborne and Mr
Davis to a new Department for Exiting the to settle matters herself”. Gove: she capriciously gave each a dress-
EU. Giving the Brexit-related jobs to Some confirmation of this came on Jan- ing down in the process. Close observers
paid-up Brexiteers insulates her from criti- uary 3rd when Sir Ivan Rogers, Britain’s say she is allergic to cutting deals and that
cisms of not supporting the policy. It also ambassador to the EU, left his job ten in cabinet she sees eye-to-eye only with
cannily reduces the chance ofa single Brex- months early. In a leaked e-mail he took ministers who, like Philip Hammond, her
iteer emerging as a rival if the process’s out- aim at “muddled thinking” on Brexit (see chancellor, and Damian Green, her wel-
come disappoints the diehard Leavers. page 35). He is not the first senior civil ser- fare secretary (and the husband of her Ox-
One minister says that, whereas the vant to leave early; Helen Bower, the re- ford tutorial partner), she has known for
cabinets of Mr Blair and Mr Brown were fu- spected chief spokeswoman at 10 Down- decades. Her sporadic attempts to lighten
rious power struggles, and Mr Cameron’s ing Street, went first. A senior minister in up are hit-and-miss: her frequent public
cabinets mostly shams, Mrs May’s cabinet the upper house, Jim O’Neill, has also mockery of Mr Johnson is making an ene-
features open discussions in which the walked out. my of him—and feels weird coming from
prime minister really listens. Another All of which is a reminder that, al- the woman who gave him his powerful
claims that she is more interested in evi- though the Labour Party’s disarray makes job in the first place.
dence than her predecessor was and Mrs May look unassailable, her position is
praises the fluency with which she shifts not entirely safe. She has a very small par- Many a conflict, many a doubt
between subjects. Acolytes insist that the liamentary majority and the Conservative There may be lessons as to Mrs May’s pos-
mighty chiefs-of-staff produce decisions Party has a knack for regicide. It looks quite sible longevity and success from her fellow
that have been properly tested (not so un- likely that the Brexit talks will founder; Mrs children of the cloth, Mr Brown and Mrs
der Mr Cameron) without prime ministeri- May insists that she wants to maintain cer- Merkel. Mr Brown, whose brief premier-
al overload (not so under Mr Brown). tain economic benefits of EU membership ship was dominated by the global finan-
Most of all, though, these arrangements but end free movement of labour, a deal cial crisis, never unified his party and was
give the prime minister what she most cov- deemed unthinkable in Brussels. That up against a strong opposition led by Mr
ets: control. Even close allies call Mrs May a could lead to economic chaos and expose Cameron. Mrs Merkel has faced crises,
control freak—and as is often the case, the her to a challenge from Mr Osborne, who is too—but for more than a decade has grown
freakery comes at the expense of trust and remaking himself as the backbench stan- through them, outwitting or co-opting her
efficiency. The “Nick and Fi” filter on poli- dard-bearer for liberal Toryism. Alterna- opposition, maintaining unquestioned su-
cies creates a bottleneck delaying urgent tively, a final deal could involve trade-offs premacy in her party.
measures (new funding to soothe the so- unpalatable to her most keenly Brexiteer Like Mrs Merkel, Mrs May has seen off
cial-care crisis was unveiled almost a MPs, who would then cut up rough. rivals through canny manoeuvring; she
month later than planned). Apparent pri- When things start to go south the defen- bides her time, knowing when to speak up
orities—like those grammar schools—have sive and needlessly belligerent tone and when (as in the referendum cam-
failed to turn into flagship policies. The shown in her tenure to date will serve her paign) to stay quiet. Like Mr Brown, she is
suggestions of workers on boards, govern- ill. For most of her end-of-term grilling by prone to overblown rhetoric, irritability
ment meddling in monetary policy and the liaison committee—a panel of MPs and indecisiveness. The biggest worry,
obligations on firms to list their foreign which scrutinises the government—she though, is that she may also share his in-
workers have all come to nothing. More re- wore an aquiline scowl, quibbling with ability to adapt—the key difference be-
grettably, so have hints of big new infra- the questions and, when pushed, cleaving tween Mr Brown and Mrs Merkel.
structure investments and house-building to evasive platitudes: “I gave the answer I Mrs May shows few signs of the ability
schemes. Westminster feels dead. to assimilate the new that has made Mrs
Comments by ministers have been dis- Merkel so successful. Her vision of leader-
owned, the Treasury feels sidelined, dip- ship, it seems, is focused on giving state-
lomats believe they are ignored. When a ments, installing processes, gathering up
consultant’s memo to the Cabinet Office information and control—and little else.
criticising Mrs May’s leadership style This makes it worryingly easy to imagine
leaked, the prime minister reportedly de- the Britain of 2018 or 2019 in disarray: her
manded that Deloitte, the firm in question, party in revolt, her ministers and partners
be “punished”. It has since withdraw from alienated, her government sclerotic, Brexit
a series of bids for government contracts, talks breaking down, the economy tanking
and ministers’ e-mails and phone records and Number10 in bunker mode.
are to be seized to prevent further leaks. For there is more to leadership than Mrs
Even the queen has reportedly grumbled May’s procedures. There is also what Peter
about Mrs May’s slogan-heavy furtiveness Hennessy, a contemporary historian, calls
about how Britain will leave the EU. “the emotional geography” of power. This
Indeed, six months after coming to means adapting to events and institutions,
power all the prime minister can say on building networks and—yes—being judi-
that subject is that “Brexit means Brexit” ciously informal sometimes: a dose of in-
and that it will be “red, white and blue” (ie stinct, a snap decision, a deal cut, a risk tak-
patriotic, rather than Caucasian, bloodied en on a wing and a prayer. It means sharing
and bruised). Her fear of losing control ex- information, accepting dissent, seeking al-
plains why, instead of holding a simple ternative opinions, staking out a position
parliamentary vote on triggering Article 50 and persuading people of it. It is this emo-
of the EU Treaty (the process by which Brit- tional landscape that Britain’s geographer
ain will leave the union), she stubbornly Very well, alone prime minister must master, if she can. 7
The Economist January 7th 2017 21
United States
Also in this section
22 Congressional ethics
23 Recruiting police officers
23 Stand-your-ground laws, measured
24 Charleston’s new museum
24 Markets for tickets
25 Lexington: Learning to love
Trumpism
swing might be thought of as a revolt of the driven inequality since 2000. A recent
3
Easing the pain lower middle. However, it was largest compendium published by the Russell
United States, median real household income* among those with incomes beneath Sage Foundation warns of growing differ-
1979=100 $30,000. Most of these voters are probably ences in wealth even among those who are
Market income: in the poorest fifth of households, though not rich. Mr Trump’s plan to reduce taxes
plus transfers plus transfers less taxes some may previously have held more lu- on capital returns and abolish them on in-
150 crative jobs. The Pew Research Centre esti- heritance could exacerbate these trends,
140 mates that the middle class, defined as much as the Reagan income-tax cuts coin-
those with incomes between two-thirds cided with growing disparity in wages.
130
and twice the median, shrank from 55% of The effect of Mr Trump’s economic poli-
120 the population in 2000 to 51% by 2014. cies on median incomes will depend on
110 whether they encourage firms to invest,
100
Reaganite or kryptonite? boosting workers’ productivity. Historical
Inequality will rise if Mr Trump succeeds evidence is not encouraging: median earn-
90 in slashing taxes for the highest earners, as ings barely grew in the 1980s. But if wages
1979 85 90 95 2000 05 10 13
it did after Ronald Reagan’s tax cuts in the continue their recent recovery, Mr Trump is
Source: CBO *Adjusted for household size
1980s. Then, the labour market was about sure to claim the credit. And, unlike his
to bifurcate into winners and losers from party, Mr Trump has shown little appetite
2 class, notes Gabriel Zucman of the Univer- globalisation and technological change. to curb spending on the middle class. A
sity of California, Berkeley. With Thomas Today, rising inequality in wealth, rather very rich elite, high poverty and plentiful
Piketty and Emmanuel Saez, two other than in wages, might be a bigger concern. government spending on the middle could
economists, Mr Zucman recently produced Mr Zucman and his co-authors find that a make Mr Trump look like a continuity can-
new estimates which harness GDP data to boom in investment income at the top has didate after all. 7
improve the familiar figures from surveys
and tax returns. They find that the incomes
Congressional ethics
of those in the 50th to 90th income percen-
tiles have grown by 40% since 1980, more
than previously thought, thanks to grow- Old bog, new tricks
ing tax exemptions. The poorer half of
WASHINGTON, DC
Americans pay roughly as much in taxes as
How to lose votes and irritate people
they receive in cash redistribution, in spite
of the EITC.
Before the financial crisis, government
redistribution kept median incomes rising
A S THEIR first major initiative of the
new year, Republican congressmen
announced a scheme so crassly self-
congressional control and limit the scope
of its investigations and its ability to
publicise its work. Paul Ryan, the Repub-
even as wages stagnated (see chart 3). Since interested as to suggest they had learned lican Speaker of the House of Repre-
then it has kept incomes flat as wages have nothing from the old one. Denizens of a sentatives, warned against this; the plan
fallen. By 2013 median household income reviled institution, and a party railroaded was nonetheless approved, by a vote of
before taxes was 1.6% lower than it was in by Donald Trump’s populist insurgency, 119 Republican congressmen to 74.
1999. But after taking off taxes and adding they planned to gut the Office of Congres- Mr Trump, no doubt aware of how
in government transfers, it was fully 13.7% sional Ethics (OCE), an independent badly this was playing, offered a mea-
higher. More recent data suggest that even investigative body designed to root out sured criticism of the congressmen’s
pre-tax incomes are now growing again: corruption. Less than 24 hours later, after initiative. In a tweet, he called the OCE
they were up by 5.2% in 2015. a hail of condemnation, they turned tail; “unfair” but suggested his Republican
The economic safety net for the poor- even so their bungling was damning. colleagues had bigger things to be getting
est, however, remains perilously thin by in- The OCE was founded by the Demo- on with. A deluge of negative comments
ternational standards. A typical jobless crats in 2008 after a run of scandals— received at their district offices made that
married couple with two children can ex- including a big one concerning the Re- point more forcefully. So did Senator
pect a welfare income, including the value publican lobbyist Jack Abramoff—high- Lindsey Graham of South Carolina, who
of food stamps, worth 23% of median pay. lighted the impunity with which some called the attempted takedown of the
The average in the OECD, a club of mostly lawmakers were abusing their office in OCE “the dumbest frickin thing I’ve ever
rich countries, is 40%. Partly as a result, rel- exchange for campaign contributions. heard”. On January 3rd, the opening day
ative poverty is higher than every other The office is empowered and equipped to of the new Congress, the plotters hastily
member of the club bar Israel. This looks investigate allegations of impropriety. It agreed to leave the OCE alone after all.
even worse as the lower-paid have borne may then report its findings to the House This delivered an easy triumph to Mr
the brunt of rising inequality. Messrs Pi- Ethics Committee and, even if that body Trump, whose tweet was credited by
ketty, Saez and Zucman find that the trend decides to take no further action, pub- many headline writers with having
since 1980 can be summarised as a shift of licise them. Anti-corruption campaigners persuaded the congressmen to change
8% of national income from the bottom consider it a bulwark against official course, albeit without much evidence. It
half of earners to the top 1%, with no effect corruption. Many congressmen consider also showed those lawmakers to lack
on those in between. it unjust and wasteful. self-awareness to an amazing degree. If
That all still leaves those whose earn- Several who have been subject to the the OCE is not working well, they should
ings place them between the middle and office’s inquiries were involved in the start a debate—in and with the public—
the poor. Median household income in effort, at a closed-door meeting of Repub- about how better to investigate and
2015 was nearly $57,000. Exit polls suggest lican congressmen, to nobble it. They prevent their abuses. To avoid an unnec-
that Mr Trump lost among voters with in- included Blake Farenthold of Texas, who essary partisan fight, they also plainly
comes beneath $50,000, as a Republican was investigated and exonerated by the need Democratic support. This is basic
presidential candidate would be expected office over an allegation of sexual harass- politics. Republican congressmen should
to. But he did much better with such voters ment. The plan was to put the OCE under really learn how to do it.
than Mitt Romney did in 2012. The positive
The Economist January 7th 2017 United States 23
Recruiting police officers Dallas, though sympathy can also boost re-
cruitment. Dallas has seen an uptick in ap-
The force is weak plications since its officers were attacked.
The last is the image of policing. The
deaths of several unarmed black men at
the hands of police officers and the ensu-
ing backlash seem to have made police
work less appealing. “We have a situation
LOS ANGELES
where law enforcement is being scruti-
Police departments struggle to stay fully staffed
nised more heavily,” says Mr Hamilton of
CHARLESTON
Filling in the gaps in America’s history
“H AMILTON”, the hip-hop Broad-
way musical about one of Ameri-
ca’s founding fathers, has broken all sorts
Last month President Obama signed
legislation which aims to eliminate bots
and intends to slap hackers with hefty
Conservatives are working hard to reconcile their beliefs with the next president’s agenda
Other conservative grandees wonder if a dose of economic
nationalism is the price of solidifying the coalition that carried
Mr Trump to power, including blue-collar voters in the Midwest
who abandoned the Democrats in droves. They praise Mr Trump
as a patriot-pragmatist in the spirit of Lincoln or Theodore Roose-
velt. They are slower to note more recent models for Trumpism,
starting with populist-nationalist movements sweeping Europe.
Parallels with Nicolas Sarkozy, France’s centre-right president
from 2007-12 and a hyperactive corporatist, are startling. Mr Sar-
kozy denounced French carmakers for producing cars in eastern
Europe (“not justifiable”, he growled), and rushed to a steelworks
to promise workers he would save their jobs (a pledge he could
not keep).
Hugh Hewitt, a conservative talk-radio host, this month will
publish “The Fourth Way”, a book-length guide to how Trum-
pism might advance bits ofthe Reagan agenda, by promoting con-
servative judges, stronger armed forces (Mr Hewitt likes Mr
Trump’s talk of a 350-ship navy) and free enterprise (above all
rolling back “the vast and growing regulatory state”). To that he
would add a “repatriation window” for corporate profits held
abroad, and a new, voter-pleasing wave of infrastructure projects,
2 zil’s prisoners, bear the burden of manag- nationalised utilities, put party hacks in
ing and paying for the system. Evo’s economy charge and failed to invest in them. La Paz’s
They, in turn, have neither the money Bolivia water reserves reached dangerously low
nor the ideas needed to improve condi- 9
levels even before the drought took hold.
tions. Politicians and judges are more eager GDP, % increase on a year earlier Such setbacks have damaged the gov-
to lock up criminals, especially if they are 6 ernment’s prestige. In October it had to
poor and black, than they are to reduce 3
cancel a popular bonus of an extra
overcrowding. About two-fifths of Brazil’s + month’s wages paid to all workers in the
prisoners are awaiting trial rather than 0 formal sector, but only in years when GDP
–
serving sentences; university graduates, growth is more than 4.5%. Mr Morales’s re-
3
priests and others are entitled to wait in General-government lations with trade unions and social move-
budget balance, % of GDP
comfier conditions. 6 ments, which once gave him unstinting
Governments also fear that a crack- support, have been hurt by disputes over
9
down on violence in prisons will cause 2006 08 10 12 14 16*17† infrastructure projects and benefits for dis-
trouble outside them. An attempt by São †Forecast
abled people. A conflict over regulation of
Source: IMF *Estimate
Paulo’s government in 2006 to curb the pri- mining by co-operatives led to the deaths
son-based operations of the PCC set off a of four miners and the murder of a vice-
campaign ofviolence by the gang’s confed- Bolivia’s woes may cut Mr Morales minister. The Central Obrera Boliviana,
erates across the state. Hundreds died over down to size. Despite the onset of the rainy the main trade-union federation, has fallen
ten days in attacks on policemen and the season in December, many districts are still out with the government. The MAS lost
reprisals they provoked. Politicians prefer rationing water. The state-owned water control of El Alto, Mr Morales’s political
to keep the violence within prison walls. 7 company that supplies La Paz, the seat of stronghold, in regional elections in 2015.
government, and El Alto, a populous city None of that deters his allies from plot-
perched on a cliff above it, ran out of water ting to keep him in power. Many lookto the
Bolivia in November. The water level in the Inca- example of the late Hugo Chávez, Vene-
chaca reservoir, which serves parts of La zuela’s left-wing leader, who lost a referen-
For Evo, for ever Paz, was far below normal in early January.
Residents of the city queue for hours to get
dum to end term limits in 2007 only to hold
another one 14 months later, which he
deliveries by lorry. Farmers and ranchers won. Under Bolivia’s constitution, a peti-
are reporting large and growing losses. tion signed by a fifth of the electorate could
This is contributing to the slowdown of trigger a re-run of the referendum to lift the
LA PAZ
the economy, which depends largely on term limit facing Mr Morales.
Evo Morales’s supporters are looking
gas exports. Their price is linked to the Another option, suggested by the vice-
for ways around term limits
price of oil, which has halved since 2014. In president, Álvaro García, is that Mr Mo-
Japan’s elderly workers find work through Koreisha were once em-
Where are the kids? ployees of Tokyo Gas, Japan’s largest sup-
Population of Okutama and Japan
Five-year age groups, 2015, % of total
Silver lining plier of natural gas to homes. They do the
same kind of work now—reading meters
OKUTAMA, TOKYO PREFECTURE JAPAN and explaining the use of appliances to
12 9 6 3 0 3 6 9 homeowners. “They have so much experi-
95+ ence and knowledge that can be put to
90-94 good use,” says Mr Ogata.
85-89 TOKYO
80-84 They can also be cheaper. Companies
As Japan ages, so too does its workforce
75-79 often hire back retirees on non-permanent
70-74
65-69
60-64
L IKE many firms in Aichi prefecture, Ja-
pan’s manufacturing heartland, Nishiji-
max, a maker of machine tools for the car
contracts offering poorer terms than their
previous ones. Takashimaya, a depart-
ment-store chain, has introduced a perfor-
55-59
50-54 industry, is struggling to find workers. Its mance-based system for such employees
45-49 solution in a country with a drum-tight la- aged 60-65 (at no extra cost to the company,
40-44 bour market is one that is increasingly it says).
35-39 common in Japan: raising the age of retire- Japan’s labour crunch has created a
30-34
25-29
ment. More than 30 of the company’s 140 chronic shortage of nursing care for elderly
20-24 employees are over 60; the oldest is 82. Put- people who are no longer fit enough to
15-19 ting qualified people out to pasture early is work. McKinsey, a consultancy, says Japan
10-14 a waste, says Hiroshi Nishijima, a manag- should encourage able-bodied elderly
5-9
er; “If they want to work, they should.” people to help. If 10% of them were to take
0-4
Since peaking at over 67m in the late up such work, the country would have an
Sources: Okutama town records; Statistics Japan
1990s, Japan’s workforce has shrunk by additional 700,000 carers by 2025, it reck-
about 2m. The government says it could ons. One way of encouraging this would
2 unlikely to move permanently just to collapse to 42m by mid-century as the pop- be to give priority to those who have
make a bureaucrat’s life easier. The inter- ulation ages and shrinks. The number of worked as carers when allocating places in
net and home delivery help them cling on, foreigners inched up in 2015 to a record nursing homes, says McKinsey. It does not
points out Mr Morita. high of 2.2m, but that is far from enough to help, however, that the state pension sys-
Okutama has tried to promote agricul- fill the labour gap. Instead of opening its tem discourages some elderly people from
ture: wasabi, a spicy vegetable that is doors wider to immigrants, Japan is trying working by cutting their benefits if they
ground up and eaten with sushi, grows to make more use of its own people who earn more than a certain amount.
well there. It hopes to appeal to families by are capable of working. At Nishijimax, managers clearly want
offering free vaccinations, free school Large companies in Japan mostly set a elderly workers to stay. The company’s
lunches and free transport. None of that mandatory retirement age of 60—mainly work routine is tailored to their needs. So,
has staved off ageing and decline. So now it as a way of reducing payroll costs in a sys- too, are the canteen’s offerings—right
is touting free housing. Mr Morita esti- tem that rewards seniority. But other busi- down to the reduced-salt miso soup. 7
mates that the town has about 450 empty nesses are less stringent. About 12.6m Japa-
homes. He wants the owners to give their nese aged 60 or older now opt to keep
homes to the town government, which working, up from 8.7m in 2000. Two-thirds Alcohol in Indonesia
they might do in order to avoid property of Japan’s over-65s say they want to stay
taxes. The government will then rent the
homes to young couples, the more fecund
gainfully employed, according to a govern-
ment survey. The age of actual retirement
Dry talk
the better. If they stay for 15 years their rent for men in Japan is now close to 70, says the
will be refunded. OECD, a rich-country think-tank. In most
Although its setting, amid steep hills, is countries people typically stop working
spectacular, Okutama is not a pretty town. before the age at which they qualify for a
SEMARANG
Its houses are neither old enough to be con- state pension. Japan, where the state pen-
Debating a ban on booze
sidered beautiful nor modern enough to sion kicks in at 61 (it is due to rise to 65 by
be comfortable. Some feature post-war
wheezes like plastic siding. Still, the pros-
pect of free accommodation some two
2025), is a rare exception.
The greying of Japan’s workforce is
clearly visible. Elderly people are increas-
O NE of Indonesia’s newest brands of
beer, Prost, traces its ancestry back to
1948 when Chandra Djojonegoro, a busi-
hours’ journey from central Tokyo might ingly seen driving taxis, serving in super- nessman, started selling a “health tonic”,
tempt some young families. And in the markets and even guarding banks. Bosses known as Anggur Orang Tua, from the
meantime, Okutama has another plan. are getting older, too. Mikio Sasaki, the back of a bright-blue lorry at night markets
A building once occupied by a junior chairman of Mitsubishi Corporation, a in the coastal city of Semarang. A troupe of
high school, which closed for lack of pu- trading company, is 79. Masamoto Yashiro, dancing dwarves would pull in the punt-
pils, is becoming a language college. Jelly- the chairman and CEO of Shinsei Bank, is ers, while Djojonegoro peddled shots of
fish, an education firm with tentacles in 87. Tsuneo Watanabe, editor-in-chief of the what was, in essence, a fortified herbal
several countries, will use it to teach Japa- world’s biggest-circulation newspaper, the wine to fishermen. It kept them warm dur-
nese to young graduates from East and Yomiuri Shimbun, is a sprightly 90. ing the chilly nights in the Java Sea.
South-East Asia. It hopes to enroll 120 stu- It is inevitable that people will stay in The tonic is still sold in bottles with dis-
dents, plus staff, which ought to make a no- the workforce longer, says Ken Ogata, the tinctive labels depicting an old Chinese
table difference in a district where there are president of Koreisha, an agency that pro- man with a thick white beard. The com-
now fewer than 350 people in their 20s. vides temporary jobs exclusively to people pany that makes it now produces a vast
Some of those students might even decide over 60. He notes that the country has little range of consumer goods, and Prost beer is
they like the place, and settle down. Whis- appetite for importing workers, so it will the latest addition to its range. It is made in
per it, but this sounds a little like a more lib- have to make more use of pensioners, a $50m brewery that opened in August
eral immigration policy. 7 women and robots. Many of those who 2015, filled with shiny stainless-steel ma- 1
30 Asia The Economist January 7th 2017
2 chinery from Germany. Thomas Dosy, est Muslim population, Indonesia is re- which were shot in New Zealand’s breath-
chief executive of the subsidiary that pro- markably permissive. Night spots in Jakar- taking wilderness).
duces Prost, says that given Orang Tua’s ta, the capital, and tourist magnets such as But for every happy Chinese couple
history in the booze business it was natural the island of Bali have their raunchy sides. snuggling up for a selfie next to a tuatara
for the company to move into Indonesia’s In Semarang, Mr Dosy predicts steady there is a grumpy New Zealander who re-
$1bn-a-year beer market. growth in domestic sales of 8-9% per year, members the way things used to be—when
It will not be straightforward. Conser- buoyed by a growing number of middle- you could walk the tracks without running
vative Muslim groups have become more class tipplers. Most Indonesians, proud of into crowds at every clearing. Many locals
assertive. Only months before the brewery their tradition of tolerance, will be hoping now wonder why their taxes, as they see it,
opened, the government slapped a ban on that he is right. 7 are paying for someone else’s holiday. Mr
the sale of beer at the small shops where Sanson would seem to agree. Entry fees
most people buy their groceries. It led to a could be used to upgrade facilities such as
13% slump in sales, according to Euromon- New Zealand’s national parks cabins, car parks and trails. A varying levy
itor, a research firm. The government min- could also help reduce numbers at some of
ister who issued the decree has since been
sacked, but his ban remains in place. And
Lord of the the popular locations by making it cheaper
to use lesser-known, but no less beautiful,
Muslim parties in parliament are still not
satisfied. They are pushing legislation that
ker-chings trails farther afield.
Some are not so sure it would work.
would ban the production, distribution Hugh Logan, a former chief of conserva-
QUEENSTOWN
and consumption of all alcoholic bever- tion for the government who now runs a
A proposal to tax users of national parks
ages. Drinkers could face two years in jail. mountaineering club, worries it would
has aroused fierce argument
The law is unlikely to pass. Muslim par- cost too much to employ staff to take mon-
ties control less than one-third of the legis-
lature’s seats. The government is propos-
ing a far more limited law aimed at curbing
N EW ZEALAND’S chiefconservation of-
ficer, Lou Sanson, caused a stir in Octo-
ber by suggesting that it might be time to
ey from hikers at entrances. It would also
be difficult to prevent tourists from sneak-
ing around the toll booths.
the production of toxic home-brews, start charging tourists for using the coun- Some argue that it would be easier to
known as oplosan, which are responsible try’s wilderness trails. New Zealanders are charge visitors a “conservation tax” when
for nearly all alcohol-related deaths in In- keen fans of their national parks. Many they enter the country. The Green Party, the
donesia. Turning Indonesia dry would be would be outraged at having to pay. But third-largest in parliament, says that add-
seen by many people as an affront to the many also worry about a huge influx of ing around NZ$18 ($12.50) to existing bor-
cultural diversity of the sprawling archi- foreigners who have been seeking the der taxes would still make the total
pelago, which has large Buddhist, Chris- same delights. amount levied less than visitors to arch-ri-
tian and Hindu minorities, as well as many In 2016 New Zealand hosted 3.5m tour- val Australia have to pay. But some travel
Muslims who are partial to a cool one. ists from overseas; by 2022 more than 4.5m companies oppose the idea. They note that
Brewers argue that alcohol is not an im- are expected every year—about the same tourists already contribute around
port from the decadent West, as the puri- as the country’s resident population. Tou- NZ$1.1bn through the country’s 15% sales
tans often claim, but has been produced rism has overtaken dairy produce as the tax. Better, such firms say, to use foreign
and consumed in Indonesia for at least 700 biggest export, helped by a surge in the tourists’ contribution to this tax for the
years. “It is part of the culture of Indone- number ofvisitors from China. The nation- maintenance of the parks.
sia,” says Michael Chin, chief executive of al parks, which make up about one-third Among the fiercest critics ofa charge are
Multi Bintang, the country’s biggest brew- of the territory, are a huge draw. About half those who point out that unfettered access
er. Indonesians consume less than one litre of the foreign tourists visit one. They are to wilderness areas is an important princi-
of alcohol per head a year, belying Muslim keen to experience the natural beauty ple for New Zealanders. It is enshrined in a
groups’ claims that booze is creating a promised by the country’s “100% Pure National Parks Act which inspires almost
health crisis. Still, even without a national New Zealand” advertising campaign (and constitution-like devotion among the
prohibition, Islamists will push for local shown off in the film adaptations of “The country’s nature-lovers. Mr Sanson has a
bans—such as the one in force in Aceh since Lord of the Rings” and “The Hobbit”, rocky path ahead. 7
2005 and adopted elsewhere.
Beyond booze, the state-backed council
of clerics, the Indonesian Ulema Council
(MUI), has in recent years passed edicts
condemning everything from homosexual
partnerships to the wearing of Santa hats.
Although these have no legal force under
Indonesia’s secular constitution, vigilantes
have sometimes used the edicts to target
revellers as well as religious and sexual mi-
norities. Partly at the MUI’s urging, parlia-
ment has passed sweeping anti-pornogra-
phy laws, which some Indonesians see as
a threat to artistic and cultural liberties.
Muslim groups are petitioning the courts
to interpret the law in a way that would
criminalise extramarital sex. They are also
making more use of laws against blasphe-
my—notably in the trial against the gover-
nor of Jakarta, Basuki Tjahaja Purnama, a
Christian of Chinese descent.
Still, for a country with the world’s larg- Welcome to Orcland
The Economist January 7th 2017 Asia 31
An authoritarian prime minister looks more secure than ever. Looks can deceive
with sedition has shot up. As for1MDB, the only conviction in Ma-
laysia related to it has been of a whistle-blowing legislator who
highlighted alleged wrongdoing by the fund’s managers.
Now perhaps Mr Najib feels that the chief risks from 1MDB are
behind him. Bear in mind that among the most assiduous investi-
gations to date have been those by America’s Department of Jus-
tice, which claims $3.5bn is missing from the fund. Yet the next
American president, Donald Trump, speaks admiringly ofMr Na-
jib, a golfing buddy. It might be hard for the department to pursue
a full-throttle investigation if Mr Trump expressed displeasure.
At any rate, the prime minister is at work covering his domes-
tic bases, including wooing the Islamist party, the Pan-Malaysian
Islamic Party (PAS). Some analysts mock the PAS leader, Abdul
Hadi Awang, as having ayatollah-like aspirations: the party has
long urged for sharia punishments to apply much more widely to
the Malay Muslims who make up nearly two-thirds of the popu-
lation. Such a proposal is not only morally but also constitution-
ally iffy. Undaunted, Mr Najib took the extraordinary step last
year of backing Mr Hadi’s private member’s bill, which aims to
increase the power of Islamic courts. With little discussion in cab-
inet or with the other 12 coalition members, the government sub-
2 ram home the importance of obedience, rule, his successor will be someone who speare’s 37), and is not as quotable. But no
Mr Xi recently held what he called a joins the Standing Committee right after matter. The timing was perfect. Tang died
“democratic life session” at which Polit- the coming congress. in 1616, the same year as Shashibiya, as
buro members read out Mao-era-style self- But there is widespread speculation Shakespeare is called in Chinese. President
criticisms as well as professions of loyalty that Mr Xi might seek to stay on in some ca- Xi Jinping described Tang as the “Shake-
to Mr Xi as the “core” leader (as the party pacity when his term ends in 2022. He speare of the East” during a state visit to
decided last October to call him). might, for instance, retire as state president Britain in 2015. The Ministry of Culture lat-
By August, when Mr Xi and his col- (for which post there is a clear two-term er organised a Tang-themed exhibition,
leagues hold an annual retreat at a beach limit) but continue as party general-secre- comparing his life and works to those of
resort near Beijing, the initial lists of lead- tary. He faces a trade-off. The more he Shakespeare. It has shown this in more
ers will be ready. Probably in October, the breaks with precedent, the longer he will than 20 countries, from Mexico to France.
Central Committee will hold its last meet- retain power—but the more personalised The two playwrights would not have
ing before the congress to approve its docu- and therefore more unstable the political heard of each other: contacts between Chi-
ments. The “19th Big” will start soon after, system itself may become. Trying to square na and Europe were rare at the time. But
and will last for about a week. The first that circle will be Mr Xi’s biggest challenge that has not deterred China’s cultural com-
meeting of the new Central Committee in the politicking of the year ahead. 7 missars from trying to weave a common
will take place the next day, followed im- narrative. A Chinese opera company
mediately by the unveiling before the press created “Coriolanus and Du Liniang”, in
of Mr Xi’s new lineup (no questions al- Literature which Shakespeare’s Roman general en-
lowed, if officials stick to precedent). counters an aristocratic lady from Tang’s
The process is cumbersome and elabo-
rate, but over the past 20 years it has pro-
There is flattery in best-known play, “The Peony Pavilion”.
The musical debuted in London, then trav-
duced remarkably stable transfers of pow-
er for a party previously prone to turbulent
friendship elled to Paris and Frankfurt. Last month
Xinhua, an official news agency, released
ones. This has been helped by the intro- an animated music-video, “When Shake-
SHANGHAI
duction of unwritten rules: a limit of two speare meets Tang Xianzu”. Its lines, set bi-
Officials are using Shakespeare to
terms for the post of general secretary, and zarrely to a rap tune, include: “You tell love
promote a bard of China’s own
compulsory retirement for Politburo mem- with English letters, I use Chinese ink to de-
bers if they are 68 or over at the time of a
congress. Mr Xi, however, is widely be-
lieved to be impatient with these restric-
L IKE many countries, China had a busy
schedule of Shakespeare-themed cele-
brations in 2016, 400 years after his death.
pict Eastern romance.”
The anniversary of Shakespeare’s
death is now over, but officially inspired
tions. He has ignored the party’s hallowed There were plays, lectures and even plans adulation of Tang carries on (a musical
notion of “collective leadership”, by accru- announced for the rebuilding of his home- about him premiered in September in Fu-
ing more power to himself than his post- town, Stratford-upon-Avon, at Sanweng- zhou, his birthplace—see picture). Chinese
Mao predecessors did. upon-Min in Jiangxi province. But as many media say that a recent hit song, “The New
If precedent is adhered to, five of the organisers saw it, Shakespeare was just an Peony Pavilion”, is likely to be performed
seven members of the Politburo’s Standing excuse. Their main aim was to use the Eng- at the end of this month on state televi-
Committee, six of its other members and lish bard to promote one of their own: Tang sion’s annual gala which is broadcast on
four of the 11 members of the party’s Cen- Xianzu. Whatever the West can do, their the eve of the lunar new year. It is often de-
tral Military Commission (as the army message was, China can do at least as well. scribed as the world’s most-watched tele-
council is known) will all start drawing Tang is well known in China, though vision programme. Officials want to culti-
their pensions. In addition, roughly half even in his home country he does not en- vate pride in Chinese literature, and boost
the 200-odd full members of the Central joy anything like the literary status of his foreign awareness of it. It is part of what
Committee (its other members, known as English counterpart—he wrote far fewer they like to call China’s “soft power”.
alternates, do not have voting rights) will works (four plays, compared with Shake- Shakespeare’s works only began to take
retire, or will have been arrested during Mr root in China after Britain defeated the
Xi’s anti-corruption campaign. This would Qing empire in the first Opium War of
make the political turnover at this year’s 1839-42. They were slow to spread. After
gatherings the biggest for decades, akin to the dynasty’s collapse in the early 20th
changing half the members of the House century, Chinese reformers viewed the
of Representatives and three-quarters of lack of a complete translation of his works
the cabinet. as humiliating. Mao was less keen on him.
Until late in 2016 there was little to sug- During his rule, Shakespeare’s works were
gest any deviation from the informal rules. banned as “capitalist poisonous weeds”.
But in October Deng Maosheng, a director Since then, however, his popularity has
of the party’s Central Policy Research Of- surged in tandem with the country’s grow-
fice, dropped a bombshell by calling the ing engagement with the West.
party’s system of retirement ages “folk- Cong Cong, co-director of a recently
lore”—a custom, not a regulation. opened Shakespeare Centre at Nanjing
The deliberate raising of doubts about University, worries that without a push by
retirement ages has triggered a round of ru- the government, Tang might slip back into
mour and concern in Beijing that Mr Xi relative obscurity. But Ms Cong says the
may be considering going further. The “Shakespeare of the East” label does Tang a
main focus is his own role. Mr Xi is in the disservice by implying that Shakespeare is
middle ofhis assumed-to-be ten-year term. the gold standard for literature. Tang
By institutional tradition, any party leader worked in a very different cultural envi-
must have served at least five years in the ronment. That makes it difficult to compare
Standing Committee before getting the top the two directly, she says. Officials, how-
job. So if Mr Xi is to abide by the ten-year The balcony scene of the East ever, will surely keep trying. 7
34 The Economist January 7th 2017
Britain
Also in this section
35 Brexit preparations
35 International development
36 Bagehot: Pierogi and integration
2 the recent rises in certain crimes. Although Cameron settled instead for a four-year
International development
car theft in general has been falling, sophis- freeze on in-work benefits for EU migrants.
ticated thefts of expensive cars by skilled
criminals have increased. Swiping posh
Brexiteers claim that, without Sir Ivan’s ex-
cessive caution, Britain could have got a lot A stingy new year
vehicles for resale and export is more diffi- more. Yet the EU’s attachment to free
cult than nicking them off the street for joy- movement is genuine and deep—even the
The juicy aid budget sparks jealousy
riding. Thieves are pinching car keys rather benefits change that Mr Cameron won
than simply breaking into vehicles, or un-
locking them remotely by hacking into
their security systems.
took 48 hours of hard pounding to secure.
Uncertainty clouds Brexit, even after
the speedy replacement of Sir Ivan by Sir
I N 2015 Britain gave away £12.1bn
($18.5bn) in foreign aid, more than any
country bar America. It was one of just
That such professionals are responsible Tim Barrow, previously ambassador to six countries to meet the UN’s target of
hints at why crime rates may have further Russia. Sir Ivan’s letter makes clear that the spending 0.7% of GDP on international
to fall. Studies in America, where crime has government has no detailed exit strategy assistance. Yet although the leaders of all
also been declining for a long time, suggest and that its negotiating team is not even Britain’s main political parties support
that men over 40 today offend at a much fully in place. Mrs May insists she will trig- this generosity, grumbles that the money
higher rate than men of that age did a cou- ger Article 50, the legal way to leave the EU, should stay at home are growing louder.
ple of decades ago. Today’s middle-aged by the end of March, earlier than Sir Ivan For the past few months newspapers
crooks learned their trade in the 1980s advised. That will set a two-year deadline have been digging up examples of exor-
when crime was relatively easy, and have for Brexit. A chunk of 2017 will be taken up bitant aid-industry salaries and alleged
carried on offending, says Mr Farrell. In by Dutch, French, German and probably mis-spending. According to the Daily
time this light-fingered generation will “re- Italian elections. Sir Ivan pointedly notes Mail, £5.2m of British cash went to an
tire”, or die. With fewer novices taking that serious multilateral negotiating expe- Ethiopian pop group (defenders point
their place, crime may dip lower still. 7 rience is in short supply in Whitehall. In out that the band was part of a project to
the EU institutions in Brussels, it is not. change attitudes about women’s roles).
His resignation supports the idea that Some backbench Conservatives have
Brexit preparations Mrs May and her ministers mistrust advis- called for aid to be redirected to pay for
ers tainted by time in Europe. Lord Mac- social care for elderly Britons. The UK
Rogers and out pherson, a former permanent secretary to
the Treasury, tweeted that, with other de-
Independence Party wants to spend it on
homeless veterans instead.
partures, it was a “wilful & total destruc- The appointment of Priti Patel as
tion of EU expertise”. Anyone with experi- head of the Department for Internation-
ence of Brussels knows it is a place in al Development in July raised hawks’
which knowledge of EU customs, laws and hopes, since she had previously called
The departure of Britain’s man in
procedures is valuable, especially after for the department to be abolished. So
Brussels lays bare a lack of Brexit plans
midnight. Mrs May could be repeatedly far, though, Ms Patel has done little to
Finding a voice
2017, as written by:
George Clooney
Sadiq Khan
Martin Sorrell
Justin Trudeau
Finding a voice
Computers have got much better at translation, voice recognition and speech synthesis, says Lane
Greene. But they still don’t understand the meaning of language
“
I
’M SORRY, Dave. I’m afraid I can’t do that.” to handle the unexpected are still far off. Artificial-in-
With chilling calm, HAL 9000, the on- telligence (AI) researchers can only laugh when asked
board computer in “2001: A Space Odys- about the prospect of an intelligent HAL, Terminator or ALSO IN THIS TQ
sey”, refuses to open the doors to Dave Rosie (the sassy robot housekeeper in “The Jetsons”). SPEECH RECOGNITION
Bowman, an astronaut who had ventured Yet although language technologies are nowhere near I hear you
outside the ship. HAL’s decision to turn on ready to replace human beings, except in a few highly
his human companion reflected a wave of fear about routine tasks, they are at last about to become good SYNTHETIC SPEECH
intelligent computers. enough to be taken seriously. They can help people Hasta la vista,robot
When the film came out in 1968, computers that spend more time doing interesting things that only hu- voice
could have proper conversations with humans mans can do. After six decades of work, much of it
seemed nearly as far away as manned flight to Jupiter. with disappointing outcomes, the past few years have MACHINE TRANSLATION
Since then, humankind has progressed quite a lot far- produced results much closer to what early pioneers Beyond Babel
ther with building machines that it can talk to, and that had hoped for.
can respond with something resembling natural Speech recognition has made remarkable ad- MEANING AND MACHINE
INTELLIGENCE
speech. Even so, communication remains difficult. If vances. Machine translation, too, has gone from terri- What are you talking
“2001” had been made to reflect the state of today’s ble to usable for getting the gist of a text, and may soon about?
language technology, the conversation might have be good enough to require only modest editing by hu-
gone something like this: “Open the pod bay doors, mans. Computerised personal assistants, such as Ap- BRAIN SCAN
Hal.” “I’m sorry, Dave. I didn’t understand the ques- ple’s Siri, Amazon’s Alexa, Google Now and Micro- Terry Winograd
tion.” “Open the pod bay doors, Hal.” “I have a list of soft’s Cortana, can now take a wide variety of
eBay results about pod doors, Dave.” questions, structured in many different ways, and re- LOOKING AHEAD
Creative and truly conversational computers able turn accurate and useful answers in a natural-sound-1 For my next trick
2 ing voice. Alexa can even respond to a request to “tell me a joke”, Many early approaches to language
but only by calling upon a database of corny quips. Computers technology—and particularly translation—
lack a sense of humour. Many early got stuck in a conceptual cul-de-sac: the
When Apple introduced Siri in 2011 it was frustrating to use, so rules-based approach. In translation, this
many people gave up. Only around a third of smartphone owners approaches meant trying to write rules to analyse the
use their personal assistants regularly, even though 95% have tried
them at some point, according to Creative Strategies, a consultan-
to language text of a sentence in the language of origin,
breaking it down into a sort of abstract
cy. Many of those discouraged users may not realise how much
they have improved.
technology “interlanguage” and rebuilding it accord-
ing to the rules of the target language.
In 1966 John Pierce was working at Bell Labs, the research arm got stuck in These approaches showed early promise.
of America’s telephone monopoly. Having overseen the team that But language is riddled with ambiguities
had built the first transistor and the first communications satellite, a conceptual and exceptions, so such systems were
he enjoyed a sterling reputation, so he was asked to take charge of hugely complicated and easily broke
a report on the state of automatic language processing for the Na- cul-de-sac down when tested on sentences beyond
tional Academy of Sciences. In the period leading up to this, schol- the simple set they had been designed for.
ars had been promising automatic translation between languages Nearly all language technologies began to get a lot better with
within a few years. the application of statistical methods, often called a “brute force”
But the report was scathing. Reviewing almost a decade of approach. This relies on software scouring vast amounts of data,
work on machine translation and automatic speech recognition, it looking for patterns and learning from precedent. For example, in
concluded that the time had come to spend money “hard-head- parsing language (breaking it down into its grammatical compo-
edly toward important, realistic and relatively short-range goals”— nents), the software learns from large bodies of text that have al-
another way of saying that language-technology research had ready been parsed by humans. It uses what it has learned to make
overpromised and underdelivered. In 1969 Pierce wrote that both its best guess about a previously unseen text. In machine transla-
the funders and eager researchers had often fooled themselves, tion, the software scans millions of words already translated by
and that “no simple, clear, sure knowledge is gained.” After that, humans, again looking for patterns. In speech recognition, the
America’s government largely closed the money tap, and research software learns from a body of recordings and the transcriptions
on language technology went into hibernation for two decades. made by humans.
The story of how it emerged from that hibernation is both sal- Thanks to the growing power of processors, falling prices for
utary and surprisingly workaday, says Mark Liberman. As profes- data storage and, most crucially, the explosion in available data,
sor of linguistics at the University of Pennsylvania and head of the this approach eventually bore fruit. Mathematical techniques that
Linguistic Data Consortium, a huge trove of texts and recordings had been known for decades came into their own, and big compa-
of human language, he knows a thing or two about the history of nies with access to enormous amounts of data were poised to
language technology. In the bad old days researchers kept their benefit. People who had been put off by the hilariously inappro-
methods in the dark and described their results in ways that were priate translations offered by online tools like BabelFish began to
hard to evaluate. But beginning in the 1980s, Charles Wayne, then have more faith in Google Translate. Apple persuaded millions of
at America’s Defence Advanced Research Projects Agency, encour- iPhone users to talk not only on their phones but to them.
aged them to try another approach: the “common task”. The final advance, which began only about five years ago,
came with the advent of deep learning through digital neural net-
Step by step works (DNNs). These are often touted as having qualities similar
Researchers would agree on a common set of practices, whether to those of the human brain: “neurons” are connected in software,
they were trying to teach computers speech recognition, speaker and connections can become stronger or weaker in the process of
identification, sentiment analysis of texts, grammatical break- learning. But Nils Lenke, head of research for Nuance, a language-
down, language identification, handwriting recognition or any- technology company, explains matter-of-factly that “DNNs are
thing else. They would set out the metrics they were aiming to im- just another kind of mathematical model,” the basis of which had
prove on, share the data sets used to train their software and allow been well understood for decades. What changed was the hard-
their results to be tested by neutral outsiders. That made the pro- ware being used.
cess far more transparent. Funding started up again and language Almost by chance, DNN researchers discovered that the graphi-
technologies began to improve, though very slowly. cal processing units (GPUs) used to render graphics fluidly in ap-
plications like video games were also bril-
liant at handling neural networks. In
Now I understand computer graphics, basic small shapes
A history of language technologies
Microsoft speech-recognition
move according to fairly simple rules, but
Scientists from John Pierce’s highly Dawn of “common system reaches human parity there are lots of shapes and many rules, re-
IBM and critical report on task” method. quiring vast numbers of simple calcula-
Georgetown language technologies Researchers share tions. The same GPUs are used to fine-tune
demonstrate published. Funding data, agree on the weights assigned to “neurons” in DNNs
a limited languishes for decades common methods
machine- of evaluation Google releases neural-net machine as they scour data to learn. The technique
translation translation for eight language pairs has already produced big leaps in quality
system “2001: A Space Odyssey”
released for all kinds of deep learning, including de-
Siri debuts on iPhone
ciphering handwriting, recognising faces
“Hey Siri” and classifying images. Now they are help-
No US government ing to improve all manner of language
research funding Statistics-based version of
for machine Google Translate launched technologies, often bringing enhance-
translation ments of up to 30%. That has shifted lan-
or speech guage technology from usable at a pinch to
recognition really rather good. But so far no one has
Source: The Economist
Speech recognition
I hear you
Computers have made huge strides in recognising human
speech
W
HEN a person speaks, air is forced out through the
lungs, making the vocal chords vibrate, which
sends out characteristic wave patterns through the
air. The features of the sounds depend on the ar-
rangement of the vocal organs, especially the ton-
gue and the lips, and the characteristic nature of the
sounds comes from peaks of energy in certain frequencies. The
vowels have frequencies called “formants”, two of which are usu-
ally enough to differentiate one vowel from another. For example,
the vowel in the English word “fleece” has its first two formants at
around 300Hz and 3,000Hz. Consonants have their own charac-
teristic features.
In principle, it should be easy to turn this stream of sound into
transcribed speech. As in other language technologies, machines
that recognise speech are trained on data gathered earlier. In this
instance, the training data are sound recordings transcribed to text
by humans, so that the software has both a sound and a text input.
All it has to do is match the two. It gets better and better at working
out how to transcribe a given chunk of sound in the same way as
humans did in the training data. The traditional matching ap-
proach was a statistical technique called a hidden Markov model
(HMM), making guesses based on what was done before. More re-
cently speech recognition has also gained from deep learning.
English has about 44 “phonemes”, the units that make up the
sound system of a language. P and b are different phonemes, be-
cause they distinguish words like pat and bat. But in English p with
a puff of air, as in “party”, and p without a puff of air, as in “spin”,
are not different phonemes, though they are in other languages. If At the level of phonemes, each language has strings that are
a computer hears the phonemes s, p, i and n back to back, it should permitted (in English, a word may begin with str-, for example) or
be able to recognise the word “spin”. banned (an English word cannot start with tsr-). The same goes for
But the nature of live speech makes this difficult for machines. words. Some strings of words are more common than others. For
Sounds are not pronounced individually, one phoneme after the example, “the” is far more likely to be followed by a noun or an ad-
other; they mostly come in a constant stream, and finding the jective than by a verb or an adverb. In making guesses about ho-
boundaries is not easy. Phonemes also differ according to the con- mophones, the computer will have remembered that in its train-
text. (Compare the l sound at the beginning of “light” with that at ing data the phrase “the right to bear arms” came up much more
the end of “full”.) Speakers differ in timbre and pitch of voice, and often than “the right to bare arms”, and will thus have made the
in accent. Conversation is far less clear than careful dictation. Peo- right guess.
ple stop and restart much more often than they realise. Training on a specific speaker greatly cuts down on the soft-
All the same, technology has gradually mitigated many of ware’s guesswork. Just a few minutes of reading training text into
these problems, so error rates in speech-recognition software have software like Dragon Dictate, made by Nuance, produces a big
fallen steadily over the years—and then sharply with the introduc- jump in accuracy. For those willing to train the software for longer,
tion of deep learning. Microphones have got better and cheaper. the improvement continues to something close to 99% accuracy
With ubiquitous wireless internet, speech recordings can easily be (meaning that of each hundred words of text, not more than one is
beamed to computers in the cloud for analysis, and even smart- wrongly added, omitted or changed). A good microphone and a
phones now often have computers powerful enough to carry out quiet room help.
this task. Advance knowledge of what kinds of things the speaker might
be talking about also increases accuracy. Words like “phlebitis”
Bear arms or bare arms? and “gastrointestinal” are not common in general discourse, and
Perhaps the most important feature of a speech-recognition sys- uncommon words are ranked lower in the probability tables the
tem is its set ofexpectations about what someone is likely to say, or software uses to guess what it has heard. But these words are com-
its “language model”. Like other training data, the language mod- mon in medicine, so creating software trained to look out for such
els are based on large amounts of real human speech, transcribed words considerably improves the result. This can be done by feed-
into text. When a speech-recognition system “hears” a stream of ing the system a large number of documents written by the speak-
sound, it makes a number of guesses about what has been said, er whose voice is to be recognised; common words and phrases
then calculates the odds that it has found the right one, based on can be extracted to improve the system’s guesses.
the kinds of words, phrases and clauses it has seen earlier in the As with all other areas of language technology, deep learning
training text. has sharply brought down error rates. In October Microsoft an-1
The Economist January 7th 2017 5
TECHNOLOGY QUARTERLY Language
Machine translation
Speak easy
Beyond Babel Human scorers’ rating* of Google Translate and human translation
Translation method Phrase-based†
(2007)
Neural-network†
(2016)
Human
3 4 5 Perfect translation= 6
Spanish
English French
Computer translations have got strikingly better, but still Chinese
I
French English
N “STAR TREK” it was a hand-held Universal Translator; in
Chinese English
“The Hitchhiker’s Guide to the Galaxy” it was the Babel
Fish popped conveniently into the ear. In science fiction, Input sentence
the meeting ofdistant civilisations generally requires some
kind of device to allow them to talk. High-quality automat- Pour l’ancienne secrétaire d’Etat, il s’agit de faire oublier un mois de
cafouillages et de convaincre l’auditoire que M. Trump n’a pas l’étoffe
ed translation seems even more magical than other kinds d’un président
of language technology because many humans struggle to speak
more than one language, let alone translate from one to another. Phrase-based† Neural-network† Human
The idea has been around since the 1950s, and computerised For the former For the former secretary The former secretary of
translation is still known by the quaint moniker “machine transla- secretary of state, this of state, it is a question state has to put behind
is to forget a month of of forgetting a month of her a month of setbacks
tion” (MT). It goes back to the early days of the cold war, when bungling and convince muddles and convincing and convince the
American scientists were trying to get computers to translate from the audience that Mr the audience that Mr audience that Mr Trump
Russian. They were inspired by the code-breaking successes of the Trump has not the Trump does not have the does not have what it
makings of a president stuff of a president takes to be a president
second world war, which had led to the development of comput-
ers in the first place. To them, a scramble of Cyrillic letters on a Source: Google *0=completely nonsense translation, 6=perfect translation †Machine translation
page of Russian text was just a coded version of English, and turn-
ing it into English was just a question of breaking the code.
Scientists at IBM and Georgetown University were among proach that would revive optimism about MT. Its Candide system
those who thought that the problem would be cracked quickly. was the first serious attempt to use statistical probabilities rather
Having programmed just six rules and a vocabulary of 250 words than rules devised by humans for translation. Statistical, “phrase-
into a computer, they gave a demonstration in New York on Janu- based” machine translation, like speech recognition, needed
ary 7th 1954 and proudly produced 60 automated translations, in- training data to learn from. Candide used Canada’s Hansard,
cluding that of “Mi pyeryedayem mislyi posryedstvom ryechyi,” which publishes that country’s parliamentary debates in French
which came out correctly as “We transmit thoughts by means of and English, providing a huge amount of data for that time. The
speech.” Leon Dostert of Georgetown, the lead scientist, breezily phrase-based approach would ensure that the translation of a
predicted that fully realised MT would be “an accomplished fact” word would take the surrounding words properly into account.
in three to five years. But quality did not take a leap until Google, which had set itself
Instead, after more than a decade of work, the report in 1966 by the goal of indexing the entire internet, decided to use those data
a committee chaired by John Pierce, mentioned in the introduc- to train its translation engines; in 2007 it switched from a rules-
tion to this report, recorded bitter disappointment with the results based engine (provided by Systran) to its own statistics-based sys-
and urged researchers to focus on narrow, achievable goals such tem. To build it, Google trawled about a trillion web pages, looking
as automated dictionaries. Government-sponsored work on MT for any text that seemed to be a translation of another—for exam-
went into near-hibernation for two decades. What little was done ple, pages designed identically but with different words, and per-
was carried out by private companies. The most notable of them haps a hint such as the address of one page ending in /en and the
was Systran, which provided rough translations, mostly to Ameri- other ending in /fr. According to Macduff Hughes, chief engineer
ca’s armed forces. on Google Translate, a simple approach using vast amounts of
data seemed more promising than a clever one with fewer data.
La plume de mon ordinateur Training on parallel texts (which linguists call corpora, the plu-
The scientists got bogged down by their rules-based approach. ral of corpus) creates a “translation model” that generates not one
Having done relatively well with their six-rule system, they came but a series ofpossible translations in the target language. The next
to believe that if they programmed in more rules, the system step is running these possibilities through a monolingual lan-
would become more sophisticated and subtle. Instead, it became guage model in the target language. This is, in effect, a set of expec-
more likely to produce nonsense. Adding extra rules, in the mod- tations about what a well-formed and typical sentence in the tar-
ern parlance of software developers, did not “scale”. get language is likely to be. Single-language models are not too
Besides the difficulty of programming grammar’s many rules hard to build. (Parallel human-translated corpora are hard to come
and exceptions, some early observers noted a conceptual pro- by; large amounts of monolingual training data are not.) As with
blem. The meaning of a word often depends not just on its dictio- the translation model, the language model uses a brute-force sta-
nary definition and the grammatical context but the meaning of tistical approach to learn from the training data, then ranks the
the rest ofthe sentence. Yehoshua Bar-Hillel, an Israeli MT pioneer, outputs from the translation model in order of plausibility.
realised that “the pen is in the box” and “the box is in the pen” Statistical machine translation rekindled optimism in the field.
would require different translations for “pen”: any pen big enough Internet users quickly discovered that Google Translate was far
to hold a box would have to be an animal enclosure, not a writing better than the rules-based online engines they had used before,
instrument. such as BabelFish. Such systems still make mistakes—sometimes
How could machines be taught enough rules to make this kind minor, sometimes hilarious, sometimes so serious or so many as
of distinction? They would have to be provided with some knowl- to make nonsense of the result. And language pairs like Chinese-
edge of the real world, a task far beyond the machines or their pro- English, which are unrelated and structurally quite different,
grammers at the time. Two decades later, IBM stumbled on an ap- make accurate translation harder than pairs of related languages1
The Economist January 7th 2017 7
TECHNOLOGY QUARTERLY Language
2 like English and German. But more often than not, Google Trans- Neural-network translation requires heavy-duty computing
late and its free online competitors, such as Microsoft’s Bing Trans- power, both for the original training of the system and in use. The
lator, offer a usable approximation. heart of such a system can be the GPUs that made the deep-learn-
Such systems are set to get better, again with the help of deep ing revolution possible, or specialised hardware like Google’s Ten-
learning from digital neural networks. The Association for Com- sor Processing Units (TPUs). Smaller translation companies and re-
putational Linguistics has been holding workshops on MT every searchers usually rent this kind of processing power in the cloud.
summer since 2006. One of the events is a competition between But the data sets used in neural-network training do not need to be
MT engines turned loose on a collection of news text. In August as extensive as those for phrase-based systems, which should give
2016, in Berlin, neural-net-based MT systems were the top per- smaller outfits a chance to compete with giants like Google.
formers (out of102), a first. Fully automated, high-quality machine translation is still a
Now Google has released its own neural-net-based engine for long way off. For now, several problems remain. All current mach-
eight language pairs, closing much of the quality gap between its ine translations proceed sentence by sentence. If the translation of
old system and a human translator. This is especially true for such a sentence depends on the meaning of earlier ones, automat-
closely related languages (like the big European ones) with lots of ed systems will make mistakes. Long sentences, despite tricks like
available training data. The results are still distinctly imperfect, but the attention model, can be hard to translate. And neural-net-
far smoother and more accurate than before. Translations be- based systems in particular struggle with rare words.
tween English and (say) Chinese and Korean are not as good yet, Training data, too, are scarce for many language pairs. They are
but the neural system has brought a clear improvement here too. plentiful between European languages, since the European Un-
ion’s institutions churn out vast amounts of material translated by
The Coca-Cola factor humans between the EU’s 24 official languages. But for smaller
Neural-network-based translation actually uses two networks. languages such resources are thin on the ground. For example,
One is an encoder. Each word of an input sentence is converted there are few Greek-Urdu parallel texts available on which to train
into a multidimensional vector (a series of numerical values), and a translation engine. So a system that claims to offer such transla-
the encoding of each new word takes into account what has hap- tion is in fact usually running it through a bridging language, near-
pened earlier in the sentence. Marcello Federico of Italy’s Fonda- ly always English. That involves two translations rather than one,
zione Bruno Kessler, a private research organisation, uses an in- multiplying the chance of errors.
triguing analogy to compare neural-net translation with the Even if machine translation is not yet perfect, technology can
phrase-based kind. The latter, he says, is like describing Coca-Cola already help humans translate much more quickly and accurately.
in terms of sugar, water, caffeine and other ingredients. By con- “Translation memories”, software that stores already translated
trast, the former encodes features such as liquidness, darkness, words and segments, first came into use as early as the 1980s. For
sweetness and fizziness. someone who frequently translates the same kind of material
Once the source sentence is encoded, a decoder network gen- (such as instruction manuals), they serve up the bits that have al-
erates a word-for-word translation, once again taking account of ready been translated, saving lots of duplication and time.
the immediately preceding word. This can cause problems when A similar trick is to train MT engines on text dealing with a nar-
the meaning of words such as pronouns depends on words men- row real-world domain, such as medicine or the law. As software
tioned much earlier in a long sentence. This problem is mitigated techniques are refined and computers get faster, training becomes
by an “attention model”, which helps maintain focus on other easier and quicker. Free software such as Moses, developed with
words in the sentence outside the immediate context. the support of the EU and used by some of its in-house translators,
can be trained by anyone with parallel cor-
pora to hand. A specialist in medical trans-
lation, for instance, can train the system on
medical translations only, which makes
them far more accurate.
At the other end of linguistic sophistica-
tion, an MT engine can be optimised for
the shorter and simpler language people
use in speech to spew out rough but near-
instantaneous speech-to-speech transla-
tions. This is what Microsoft’s Skype Trans-
lator does. Its quality is improved by being
trained on speech (things like film subtitles
and common spoken phrases) rather than
the kind of parallel text produced by the
European Parliament.
Translation management has also ben-
efited from innovation, with clever soft-
ware allowing companies quickly to com-
bine the best of MT, translation memory,
customisation by the individual translator
and so on. Translation-management soft-
ware aims to cut out the agencies that have
been acting as middlemen between clients
and an army of freelance translators. Jack
Welde, the founder of Smartling, an indus-
try favourite, says that in future translation
customers will choose how much human
intervention is needed for a translation. A
quick automated one will do for low-1
8 The Economist January 7th 2017
TECHNOLOGY QUARTERLY Language
2 stakes content with a short life, but the cold today?” The assistants know a few things about users, such as
most important content will still require a where they live and who their family are, so they can be personal,
fully hand-crafted and edited version. Not- Computer too: “How’s my commute looking?” “Text my wife I’ll be home in
ing that MT has both determined boosters 15 minutes.”
and committed detractors, Mr Welde says translation And they get better with time. Apple’s Siri receives 2bn re-
he is neither: “If you take a dogmatic quests per week, which (after being anonymised) are used for fur-
stance, you’re not optimised for the needs is still ther teaching. For example, Apple says Siri knows every possible
of the customer.”
Translation software will go on getting
known as way that users ask about a sports score. She also has a delightful
answer for children who ask about Father Christmas. Microsoft
better. Not only will engineers keep tweak-
ing their statistical models and neural net-
“machine learned from some of its previous natural-language platforms that
about 10% of human interactions were “chitchat”, from “tell me a
works, but users themselves will make im- translation” joke” to “who’s your daddy?”, and used such chat to teach its digi-
provements to their own systems. For tal assistant, Cortana.
example, a small but much-admired startup, Lilt, uses phrase- The writing team for Cortana includes two playwrights, a poet,
based MT as the basis for a translation, but an easy-to-use interface a screenwriter and a novelist. Google hired writers from Pixar, an
allows the translator to correct and improve the MT system’s out- animated-film studio, and The Onion, a satirical newspaper, to
put. Every time this is done, the corrections are fed back into the make its new Google Assistant funnier. No wonder people often
translation engine, which learns and improves in real time. Users thank their digital helpers for a job well done. The assistants’ re-
can build several different memories—a medical one, a financial plies range from “My pleasure, as always” to “You don’t need to
one and so on—which will help with future translations in that thank me.”
specialist field.
TAUS, an industry group, recently issued a report on the state of Good at grammar
the translation industry saying that “in the past few years the How do natural-language platforms know what people want?
translation industry has burst with new tools, platforms and sol- They not only recognise the words a person uses, but break down
utions.” Earlier this year Jaap van der Meer, TAUS’s founder and di- speech for both grammar and meaning. Grammar parsing is rela-
rector, wrote a provocative blogpost entitled “The Future Does Not tively advanced; it is the domain of the well-established field of
Need Translators”, arguing that the quality of MT will keep im- “natural-language processing”. But meaning comes under the
proving, and that for many applications less-than-perfect transla- heading of “natural-language understanding”, which is far harder.
tion will be good enough. First, parsing. Most people are not very good at analysing the
The “translator” of the future is likely to be more like a quality- syntax of sentences, but computers have become quite adept at it,
control expert, deciding which texts need the most attention to de- even though most sentences are ambiguous in ways humans are
tail and editing the output of MT software. That may be necessary rarely aware of. Take a sign on a public fountain that says, “This is
because computers, no matter how sophisticated they have be- not drinking water.” Humans understand it to mean that the water
come, cannot yet truly grasp what a text means. 7 (“this”) is not a certain kind of water (“drinking water”). But a com-
puter might just as easily parse it to say that “this” (the fountain) is
not at present doing something (“drinking water”).
As sentences get longer, the number of grammatically possible
Meaning and machine intelligence but nonsensical options multiplies exponentially. How can a
machine parser know which is the right one? It helps for it to know
What are you talking that some combinations of words are more common than others:
the phrase “drinking water” is widely used, so parsers trained on
large volumes of English will rate those two words as likely to be
about? joined in a noun phrase. And some structures are more common
than others: “noun verb noun noun” may be much more common
than “noun noun verb noun”. A machine parser can compute the
Machines still cannot conduct proper conversations with overall probability of all combinations and pick the likeliest.
humans because they do not understand the world A “lexicalised” parser might do even better. Take the Groucho
Marx joke, “One morning I shot an elephant in my pyjamas. How
I
N “BLACK MIRROR”, a British science-fiction satire series he got in my pyjamas, I’ll never know.” The first sentence is ambig-
set in a dystopian near future, a young woman loses her uous (which makes the joke)—grammatically both “I” and “an ele-
boyfriend in a car accident. A friend offers to help her deal phant” can attach to the prepositional phrase “in my pyjamas”.
with her grief. The dead man was a keen social-media user, But a lexicalised parser would recognise that “I [verb phrase] in
and his archived accounts can be used to recreate his per- my pyjamas” is far more common than “elephant in my pyjamas”,
sonality. Before long she is messaging with a facsimile, and so assign that parse a higher probability.
then speaking to one. As the system learns to mimic him ever bet- But meaning is harder to pin down than syntax. “The boy
ter, he becomes increasingly real. kicked the ball” and “The ball was kicked by the boy” have the
This is not quite as bizarre as it sounds. Computers today can same meaning but a different structure. “Time flies like an arrow”
already produce an eerie echo of human language if fed with the can mean either that time flies in the way that an arrow flies, or
appropriate material. What they cannot yet do is have true conver- that insects called “time flies” are fond of an arrow.
sations. Truly robust interaction between man and machine “Who plays Thor in ‘Thor’?” Your correspondent could not re-
would require a broad understanding of the world. In the absence member the beefy Australian who played the eponymous Norse
of that, computers are not able to talk about a wide range of topics, god in the Marvel superhero film. But when he asked his iPhone,
follow long conversations or handle surprises. Siri came up with an unexpected reply: “I don’t see any movies
Machines trained to do a narrow range of tasks, though, can matching ‘Thor’ playing in Thor, IA, US, today.” Thor, Iowa, with a
perform surprisingly well. The most obvious examples are the population of 184, was thousands of miles away, and “Thor”, the
digital assistants created by the technology giants. Users can ask film, has been out of cinemas for years. Siri parsed the question
them questions in a variety of natural ways: “What’s the tempera- perfectly properly, but the reply was absurd, violating the rules of
ture in London?” “How’s the weather outside?” “Is it going to be what linguists call pragmatics: the shared knowledge and under-1
The Economist January 7th 2017 9
TECHNOLOGY QUARTERLY Language
2 standing that people use to make sense of the often messy human Fernando Pereira of Google points out why. Automated speech
language they hear. “Can you reach the salt?” is not a request for recognition and machine translation have something in common:
information but for salt. Natural-language systems have to be there are huge stores of data (recordings and transcripts for speech
manually programmed to handle such requests as humans expect recognition, parallel corpora for translation) that can be used to
them, and not literally. train machines. But there are no training data for common sense.
Knowledge of the real world is another matter. AI has helped
Multiple choice data-rich companies such as America’s West-Coast tech giants or-
Shared information is also built up over the course of a conversa- ganise much of the world’s information into interactive databases
tion, which is why digital assistants can struggle with twists and such as Google’s Knowledge Graph. Some of the content of that
turns in conversations. Tell an assistant, “I’d like to go to an Italian appears in a box to the right of a Google page of search results for a
restaurant with my wife,” and it might suggest a restaurant. But famous figure or thing. It knows that Jacob Bernoulli studied at the
then ask, “is it close to her office?”, and the assistant must grasp the University of Basel (as did other people, linked to Bernoulli
meanings of “it” (the restaurant) and “her” (the wife), which it will through this node in the Graph) and wrote “On the Law of Large
find surprisingly tricky. Nuance, the language-technology firm, Numbers” (which it knows is a book).
which provides natural-language platforms to many other com- Organising information this way is not difficult for a company
panies, is working on a “concierge” that can handle this type of with lots of data and good AI capabilities, but linking information
challenge, but it is still a prototype. to language is hard. Google touts its assistant’s ability to answer
Such a concierge must also offer only restaurants that are open. questions like “Who was president when the Rangers won the
Linking requests to common sense (knowing that no one wants to World Series?” But Mr Pereira concedes that this was the result of
be sent to a closed restaurant), as well as a knowledge of the real explicit training. Another such complex query—“What was the
world (knowing which restaurants are closed), is one of the most population of London when Samuel Johnson wrote his dictio-
difficult challenges for language technologies. nary?”—would flummox the assistant, even though the Graph
Common sense, an old observation goes, is uncommon knows about things like the historical population of London and
enough in humans. Programming it into computers is harder still. the date of Johnson’s dictionary. IBM’s Watson system, which in1
2 2011 beat two human champions at the lack of anything challenging to do would be harmful to people.
quiz show “Jeopardy!”, succeeded mainly Fortunately, the tasks that talking machines can take off hu-
by calculating huge numbers of potential mans’ to-do lists are the sort that many would happily give up.
answers based on key words by probabili- Machines are increasingly able to handle difficult but well-de-
ty, not by a human-like understanding of fined jobs. Soon all that their users will have to do is pipe up and
the question. ask them, using a naturally phrased voice command. Once upon a
Making real-world information com- time, just one tinkerer in a given family knew how to work the
putable is challenging, but it has inspired computer or the video recorder. Then graphical interfaces (icons
some creative approaches. Cortical.io, a and a mouse) and touchscreens made such technology accessible
Vienna-based startup, took hundreds of to everyone. Frank Chen of Andreessen Horowitz, a venture-capi-
Wikipedia articles, cut them into thou- tal firm, sees natural-language interfaces between humans and
sands ofsmall snippets ofinformation and machines as just another step in making information and services
ran an “unsupervised” machine-learning What available to all. Silicon Valley, he says, is enjoying a golden age of
AI technologies. Just as in the early 1990s companies were piling
algorithm over it that required the comput-
er not to look for anything in particular but machines online and building websites without quite knowing why, now
to find patterns. These patterns were then everyone is going for natural language. Yet, he adds, “we’re in 1994
represented as a visual “semantic finger- cannot yet for voice.”
print” on a grid of 128x128 pixels. Clumps 1995 will soon come. This does not mean that people will com-
of pixels in similar places represented se- do is have municate with their computers exclusively by talking to them.
mantic similarity. This method can be used
to disambiguate words with multiple
true con- Websites did not make the telephone obsolete, and mobile de-
vices did not make desktop computers obsolete. In the same way,
meanings: the fingerprint of“organ” shares
features with both “liver” and “piano” (be-
versations people will continue to have a choice between voice and text
when interacting with their machines.
cause the word occurs with both in different parts of the training Not all will choose voice. For example, in Japan yammering
data). This might allow a natural-language system to distinguish into a phone is not done in public, whether the interlocutor is a hu-
between pianos and church organs on one hand, and livers and man or a digital assistant, so usage of Siri is low during business
other internal organs on the other. hours but high in the evening and at the weekend. For others,
Proper conversation between humans and machines can be voice-enabled technology is an obvious boon. It allows dyslexic
seen as a series of linked challenges: speech recognition, speech people to write without typing, and the very elderly may find it
synthesis, syntactic analysis, semantic analysis, pragmatic under- easier to talk than to type on a tiny keyboard. The very young,
standing, dialogue, common sense and real-world knowledge. Be- some of whom today learn to type before they can write, may
cause all the technologies have to work together, the chain as a soon learn to talk to machines before they can type.
whole is only as strong as its weakest link, and the first few ofthese Those with injuries or disabilities that make it hard for them to
are far better developed than the last few. write will also benefit. Microsoft is justifiably proud of a new de-
The hardest part is linking them together. Scientists do not vice that will allow people with amyotrophic lateral sclerosis
know how the human brain draws on so many different kinds of (ALS), which immobilises nearly all of the body but leaves the
knowledge at the same time. Programming a machine to replicate mind working, to speak by using their eyes to pick letters on a
that feat is very much a work in progress. 7 screen. The critical part is predictive text, which improves as it gets
used to a particular individual. An experienced user will be able
to “speak” at around 15 words per minute.
People may even turn to machines for company. Microsoft’s
Looking ahead Xiaoice, a chatbot launched in China, learns to come up with the
responses that will keep a conversation going longest. Nobody
For my next trick would think it was human, but it does make users open up in sur-
prising ways. Jibo, a new “social robot”, is intended to tell children
stories, help far-flung relatives stay in touch and the like.
Another group that may benefit from technology is smaller
language communities. Networked computers can encourage a
winner-take-all effect: if there is a lot of good software and content
in English and Chinese, smaller languages become less valuable
Talking machines are the new must-haves online. If they are really tiny, their very survival may be at stake.
But Ross Perlin of the Endangered Languages Alliance notes that
I
N “WALL-E”, an animated children’s film set in the future, new software allows researchers to document small languages
all humankind lives on a spaceship after the Earth’s envi- more quickly than ever. With enough data comes the possibility
ronment has been trashed. The humans are whisked of developing resources—from speech recognition to interfaces
around in intelligent hovering chairs; machines take care with software—for smaller and smaller languages. The Silicon Val-
of their every need, so they are all morbidly obese. Even ley giants already localise their services in dozens of languages;
the ship’s captain is not really in charge; the actual pilot is neural networks and other software allow new versions to be gen-
an intelligent and malevolent talking robot, Auto, and like so erated faster and more efficiently than ever.
many talking machines in science fiction, he eventually makes a There are two big downsides to the rise in natural-language
grab for power. technologies: the implications for privacy, and the disruption it
Speech is quintessentially human, so it is hard to imagine ma- will bring to many jobs.
chines that can truly speak conversationally as humans do with- Increasingly, devices are always listening. Digital assistants like
out also imagining them to be superintelligent. And if they are Alexa, Cortana, Siri and Google Assistant are programmed to wait
superintelligent, with none of humans’ flaws, it is hard to imagine for a prompt, such as “Hey, Siri” or “OK, Google”, to activate them.
them not wanting to take over, not only for their good but for that But allowing always-on microphones into people’s pockets and
of humanity. Even in a fairly benevolent future like “WALL-E’s”, homes amounts to a further erosion of traditional expectations of
where the machines are doing all the work, it is easy to see that the privacy. The same might be said for all the ways in which language1
The Economist January 7th 2017 11
TECHNOLOGY QUARTERLY Language
2 software improves by training on a single user’s voice, vocabulary, officer, likes to think that this will free chief financial officers from
written documents and habits. having to write up the same old routine analyses for the board,
All the big companies’ location-based services—even the accel- giving them time to develop more creative approaches.
erometers in phones that detect small movements—are making Carl Benedikt Frey, an economist at Oxford University, has re-
ever-improving guesses about users’ wants and needs. The mo- searched the likely effect of artificial intelligence on the labour
ment when a digital assistant surprises a user with “The chemist is market and concluded that the jobs most likely to remain immune
nearby—do you want to buy more haemorrhoid cream, Steve?” include those requiring creativity and skill at complex social inter-
could be when many may choose to reassess the trade-off be- actions. But not every human has those traits. Call centres may
tween amazing new services and old-fashioned privacy. The tech need fewer people as more routine work is handled by automated
companies can help by giving users more choice; the latest iPhone systems, but the trickier inquiries will still go to humans.
will not be activated when it is laid face down on a table. But hack- Much of this seems familiar. When Google search first became
ers will inevitably find ways to get at some of these data. available, it turned up documents in seconds that would have tak-
en a human operator hours, days or years to find. This removed
Hey, Siri, find me a job much of the drudgery from being a researcher, librarian or jour-
The other big concern is for jobs. To the extent that they are rou- nalist. More recently, young lawyers and paralegals have taken to
tine, they face being automated away. A good example is customer using e-discovery. These innovations have not destroyed the pro-
support. When people contact a company for help, the initial en- fessions concerned but merely reshaped them.
counter is usually highly scripted. A company employee will veri- Machines that relieve drudgery and allow people to do more
fy a customer’s identity and follow a decision-tree. Language tech- interesting jobs are a fine thing. In net terms they may even create
nology is now mature enough to take on many of these tasks. extra jobs. But any big adjustment is most painful for those least
For a long transition period humans will still be needed, but able to adapt. Upheavals brought about by social changes—like
the work they do will become less routine. Nuance, which sells the emancipation of women or the globalisation of labour mar-
lots of automated online and phone-based help systems, is bull- kets—are already hard for some
ish on voice biometrics (customers identifying themselves by say- people to bear. When those
ing “my voice is my password”). Using around 200 parameters for changes are wrought by ma- OFFER TO READERS
identifying a speaker, it is probably more secure than a fingerprint, chines, they become even harder, Reprints of Technology Quarterly
are available at US$7.00 each,
says Brett Beranek, a senior manager at the company. It will also and all the more so when those
with a minimum of 5 copies, plus
eliminate the tedium, for both customers and support workers, of machines seem to behave more 10% postage in the United
going through multi-step identification procedures with PINs, and more like humans. People al- States, 15% postage in Mexico
passwords and security questions. When Barclays, a British bank, ready treat inanimate objects as and Canada. Add tax CA, DC, IL,
offered it to frequent users of customer-support services, 84% if they were alive: who has never NY, VA; GST in Canada.
signed up within five months. shouted at a computer in frustra-
For orders to NY, please add tax
Digital assistants on personal smartphones can get away with tion? The more that machines based on cost of reprints plus
mistakes, but for some business applications the tolerance for er- talk, and the more that they seem postage.For classroom use or
ror is close to zero, notes Nikita Ivanov. His company, Datalingvo, a to understand people, the more quantities over 50, please
Silicon Valley startup, answers questions phrased in natural lan- their users will be tempted to at- telephone for discount
guage about a company’s business data. If a user wants to know tribute human traits to them. information.
which online ads resulted in the most sales in California last That raises questions about Please send your order with
month, the software automatically translates his typed question what it means to be human. Lan- payment by cheque or money
into a database query. But behind the scenes a human working for guage is widely seen as human- order to:
Datalingvo vets the query to make sure it is correct. This is because kind’s most distinguishing trait.
Jill Kaletha of Foster Printing
the stakes are high: the technology is bound to make mistakes in AI researchers insist that their
Service
its early days, and users could make decisions based on bad data. machines do not think like peo-
This process can work the other way round, too: rather than ple, but if they can listen and talk Telephone: 866 879 9144,
natural-language input producing data, data can produce lan- like humans, what does that extension 168, or e-mail:
guage. Arria, a company based in London, makes software into make them? As humans teach jillk@fostereprinting.com
which a spreadsheet full of data can be dragged and dropped, to ever more capable machines to (American Express, Visa and
be turned automatically into a written description ofthe contents, use language, the once-obvious MasterCard accepted)
complete with trends. Matt Gould, the company’s chief strategy line between them will blur. 7
12 The Economist January 7th 2017
The Economist January 7th 2017 37
Middle East and Africa
Also in this section
38 Science amid the sheep farmers
38 Zimbabwe’s sex trade
39 The battle for Mosul continues
40 Trump and the West Bank
41 Israel’s divisions
2 ent-led school governing bodies are meant And it is everything state schools are not. Early results show that its pupils are on
to hold teachers to account, they are more Its 360 pupils begin learning at 7.30am and average a year ahead of their peers. Spark
often controlled by the union, or in some end around 3pm-4pm; most state schools runs eight schools and plans to have 20 by
cases by gangs. close at 1.30pm. At the start of the day pu- 2019. Other operators, such as Future Na-
But even if there were better oversight pils gather for mindfulness exercises, tion, co-founded by Sizwe Nxasana, a for-
most teachers would struggle to shape up. maths questions, pledges to work hard— mer banker, are also expanding. “We are
In one study in 2007 maths teachers of 11- and a blood-pumping rendition of Katy never going to have a larger footprint than
and 12-year-olds sat tests similar to those Perry’s “Firework”. “We have an emotional [the] government but we can influence it,”
taken by their class; questions included curriculum as well as an academic one,” hopes Stacey Brewer, Spark’s founder.
simple calculations of fractions and ratios. says Bailey Thomson, a Spark director. Another promising scheme is the “col-
A scandalous 79% of teachers scored be- Pupils attend maths lessons based on laboration schools” pilot in the Western
low the level expected of the pupils. The Singapore’s curriculum; literacy classes Cape, based on academies in England and
average14-year-old in Singapore and South draw on how England teaches phonics. charter schools in America. The five col-
Korea performs much better. Crucially, teachers are not members of laboration schools are funded by the state
It does not have to be this way. Spark SADTU. But they receive 250 hours of pro- but run by independent operators. In what
School Bramley in Johannesburg is a low- fessional development per year, about as Helen Zille, the premier of the Western
cost private school, spending roughly as much as the average state-school teacher Cape, calls “a seminal moment”, the par-
much per pupil as the average state school. gets in a decade. ents of Oranjekloof pupils petitioned to
keep the school in the collaboration pro-
gramme when unions tried to oppose it.
Astronomers v sheep farmers in South Africa
Ms Zille wants to open a “critical mass” of
Stars and baas collaboration schools to inject competition
into the public system.
Spark and the collaboration schools
THE KAROO
suggest that South African education need
A telescope in the desert meets NIMBYism
not be doomed. But together they account
BAGHDAD
Behind the battle for Mosul lies a struggle for power in Baghdad
Charlemagne Nailed it
2 understanding state failure is “institutions, South Sudanese—a quarter of the popula- The splintering of South Sudan can be
institutions, institutions”. The world’s tion—have fled their homes. Were it not for glimpsed in the “protection of civilians”
newest country, South Sudan has received food aid, often dropped out of planes onto camps maintained by the UN. One in Juba
billions of dollars of aid and the advice of remote villages, hundreds of thousands holds almost 40,000 people. At night, gun-
swarms of consultants since seceding from would starve. shots are common and aid workers refuse
Sudan in 2011, but has failed to build any in- South Sudan failed to build institutions to venture inside. Most of the residents are
stitutions worthy of the name. Afghani- that transcended tribal loyalties or curbed Nuers, like Mr Machar, who have been
stan faces a terrifying insurgency but has a the power of warlords. Torit, where Ms stranded here since the civil war began.
president doing his best to restore order. Mandi boarded that minibus to Kenya, is a They are sure who is to blame. “Our tribe
States are not wretched and unstable good place to observe the hollowness of was killed by the government and so we
because of geography—if so, how to ex- the country’s government. Though it is came here,” says Kikany Kuol Wuol, a com-
plain the success of landlocked Botswana? capital of one of South Sudan’s 28 states, it munity chairman in the camp. “We cannot
Nor is culture the main culprit: if so, South feels like a military outpost. Troops in leave, we have nowhere to go. If our wom-
Koreans would not be more than 20 times “technicals”—pick-up trucks with mount- en just go outside to look for firewood they
richer than North Koreans. Some societies ed machine-guns—patrol the streets. are raped.” When fighting broke out in
have “inclusive institutions that foster eco- There are plenty of government build- Juba in July between Mr Machar’s forces
nomic growth”; others have “extractive in- ings, including state ministries of educa- and the government, it spread into the
stitutions that hamper [it]”. South Sudan is tion, culture and health. But none of them camp as UN peacekeepers withdrew. The
an extreme example of the latter. does much. Teachers were last paid in Sep- problem, says Mr Kuol, is that: “This is a
tember, says Jacob Atari, the local educa- government only for Dinkas. The rest of us
Never look back tion minister. Inflation of over 800% they want to starve to death.” Everyone in
“Everybody I know is getting out,” says means their monthly salary of around 300 the camp supports Mr Machar, he says.
Joyce Mandi, as she mixes maize porridge South Sudanese pounds is now worth less Mr Acemoglu and Mr Robinson are pes-
for her six children at a bus stop in South than $4. Over 70% of children are out of simistic about failed nations’ chances of
Sudan. Around her, young men heave bags school, says Mr Atari. turning around. Extractive institutions
and mattresses onto the roof of a minibus. Nowhere in South Sudan does the state typically have historical roots. For exam-
Ms Mandi is fleeing her village and head- do what it is supposed to. Only 27% of ple, the authors trace the failure of today’s
ing for Kakuma, a refugee camp in neigh- adults can read, according to the UN. Pre- Democratic Republic of Congo partly to
bouring Kenya. Her husband has gone into ventable diseases such as cholera, measles the pre-colonial Kingdom of Kongo, where
the bush, she says, to fight the government. and malaria are rampant. The rule of law is taxes were arbitrary (one was levied
The South Sudanese, who are mostly a distant dream. whenever the king’s beret fell off) and the
black African and non-Muslim, fought for The country’s political system is in the- elite sold their subjects to European slav-
half a century to secede from Sudan. Arab ory decentralised, but in reality the money ers. Peasants therefore lived deep in the for-
Muslims from the north used to oppress flows through Juba, the national capital. est, to hide from slavers and tax-collectors.
and enslave them. Perhaps 2m southern- And instead ofbeing distributed to states, it They did not adopt new technology, such
ers died in the war of secession. But few is typically stolen or spent on weapons. as the plough, even when they heard of it.
think life has got better since then. Politics is a euphemism for armed battles Why bother, when any surplus was sub-
Those now in charge are former guerril- over plunder. The warlord who wins can ject to seizure? Modern Congolese farmers
las from the Sudan People’s Liberation steal the oil and pay his troops. (Or, he can make similar complaints.
Army (SPLA), a group of tribal militias un- simply let them rob civilians.) “Why Nations Fail” argues that “the
ited only by hatred of the north. The first The fighting becomes tribal because politics of the vast majority of societies
president, Salva Kiir, tried to hold the warlords recruit by stirring up ethnic ten- throughout history has led, and still leads
SPLA’s factions together by paying them off sion so that their kinsmen will rally to today, to extractive institutions.” These
with petrodollars (oil is almost the only them. This creates a vicious circle. Lacking tend to last because they give rulers the re-
thing South Sudan exports). Alex de Waal protection from other institutions, people sources to pay armies, bribe judges and rig
of Tufts University estimates that in 2013 seekit from their own tribe. Rather than de- elections to stay in power. These rulers
the government paid salaries for 320,000 mand evenhanded government, they back adopt bad policies not because they are ig-
soldiers, police and militiamen—more tribal leaders, knowing that they will steal norant of good ones but on purpose. Let-
than a tenth of all men aged 15-54. Many of and hoping they will share the spoils with ting your relatives embezzle is bad for the
these soldiers did not exist: their pay was their kin. nation but great for your family finances. 1
pocketed by the warlords supposedly com-
manding them.
But the state’s largesse did not buy loy- BEST: Finland 18.8
2 But failed states are not doomed to stay October he convinced them to pledge
that way. Between 2007 and 2016, accord- $15bn over the next four years. Yet he is
ing to the Fragile States Index, 91 countries leery of how aid is dispensed. The flood of
grew more stable and 70 grew shakier. foreign cash that followed the American-
Among those improving were giants such led toppling of the Taliban government in
as China, Indonesia, Mexico and Brazil. 2001 often undermined the state or was
The worst performers were mostly small- wasted. Aid agencies paid salaries 20 times
er, such as Libya and Syria. higher than the Afghan civil service,
Even states that have collapsed com- prompting the best officials to quit to work
pletely can be rebuilt. Liberia and Sierra Le- as drivers and interpreters. Mr Ghani has
one were stalked by drug-addled child sol- long argued that aid should flow through
diers a decade and a half ago; now both are the national government, rather than sup-
reasonably calm. The key is nearly always port a parallel state that can pack up and go
better leadership: think of how China when donor fashions change. He may be
changed after Mao died. Many bad rulers getting his way: roughly half of aid now
continue deliberately to adopt bad poli- passes through the national budget, a
cies, but they can be—and often are—re- share that is expected to rise.
placed with better ones. Even with a leader determined to make
good choices, building an honest state is
Instructions included hard. Mr Ghani complains of inaccurate in-
Afghanistan’s president since 2014, Mr formation. “There were three databases in
Ghani is a former academic and author of the Ministry ofEducation: one for teachers,
a book called “Fixing Failed States”. His one for salaries, one for schools…they
TED talk on fixing broken states has been weren’t talking to each other.” Faulty re-
viewed 750,000 times. Now he is trying to Politics by other means in South Sudan cords make it easier for money to vanish.
put his own theories into practice. Digital payments should help, he says—the
Yet he admits that rebuilding Afghani- NATO air power combined with Ameri- police thought they had received a pay rise
stan is more complex than he expected. can-trained Afghan special forces pack “an when the first experiments with mobile
The insurgents draw support from several offensive punch”, says General Nicholson. payments began, because commanders
sources: local grievances, tribal animos- The Taliban cannot mass troops for fear of could no longer skim their wages. But there
ities, global Islamist networks, organised NATO bombs. However, they have “safety is a long way to go.
crime (Afghanistan is the world’s largest outside the country”. “A decade ago, if you went into a minis-
producer of opium) and the Pakistani secu- Mr Ghani is less tolerant of corruption ter’s office, you’d see dust on the desk, no
rity services. In 2015 Mr Ghani accused than was his predecessor, Hamid Karzai, computer and the minister picking his toe-
Pakistan of being in an “undeclared state and appears to have cleaned up customs nails,” says a Western official in Kabul.
of hostility” towards his country. Now he and government procurement a fair bit. He “Now you have competent ministers and
goes further. “In October it was almost a has improved tax collection and promoted lots of young professional staff who keep
declared state of hostility,” he says. The Ta- infrastructure projects, such as rail links in touch via WhatsApp and speak English.
liban enjoy havens in Pakistan’s lawless ar- and power plants, in the hope that Afghan- The bad news is that Ghani is still learning
eas and, analysts suspect, direct help from istan will become a central Asian hub. (He how to be a politician. Karzai would get on
Pakistani spooks, some of whom would notes with satisfaction that the Taliban the phone with tribal leaders and chat
rather have Afghanistan in chaos than see have said they will not attack such about their fathers’ health [before talking
India gain a foothold there. Recent suicide- schemes.) He promotes education for business]. Ghani tries to book them for a
bombs in Kabul appear to have contained women, which was banned when the Tali- ten-minute meeting, and hustles them out
military-grade explosives, which Afghans ban ruled Afghanistan in 1996-2001. To of the door before the tea is cold.”
assume came from over the border. conservative Afghans who think this This is a common criticism. Mr Ghani is
Mr Ghani has a clear idea of the state’s would lead to illicit mixing with men, he good at retail politics (he won the disputed
basic functions. First, it must uphold the has a convincing response. “In the remote election in 2014 partly because he had
rule of law. Second, it must secure a mo- provinces, they are asking for women doc- spent so much time sitting in villages ask-
nopoly on the use of violence. The two are tors,” he points out. How can they have fe- ing ordinary Afghans what they wanted).
linked. As Sarah Chayes points out in male doctors if they do not allow their But he is a technocrat among warlords,
“Thieves of State”, when people see the daughters to go to school? some of whom have been made billion-
state as predatory, they are more likely to Nonetheless, a survey by the Asia Foun- aires by the drugs trade. He rules in uneasy
support insurgents. She cites the example dation finds that only 29% of Afghans be- coalition with a “chief executive” with ill-
of an Afghan who was shaken down nine lieve the country is moving in the right di- defined powers: Abdullah Abdullah, the
times by police on a single journey, and rection. This is largely because 70% fear for man he beat in 2014. His vice-president is a
vowed not to warn them if he saw the Tali- their safety—the highest level in over a de- blood-spattered warlord. The president
ban planting a bomb to kill them. cade. However, a slim majority (54%) say will struggle to build a clean state when so
Mr Ghani justly takes credit for the fact the army is getting better at providing secu- many bigwigs prefer it dirty, critics say.
that the Taliban did not overthrow the rity, while only 20% say it is getting worse. Mr Ghani dismisses the charge. “If poli-
state after Mr Obama’s pull-out. “In 2015 Public perceptions of corruption have tics becomes all tactics, where would you
we were in danger, because the global and barely budged since Mr Ghani came to produce change?” he asks. He insists that
regional consensus was that we would not power, with 89% of Afghans saying it is a he bends over backwards to be respectful
be able to hold,” he says. Now, says Gen- problem in their daily life. More encourag- of tribal leaders, “but it cannot be at the ex-
eral John Nicholson, the commander of ingly, the share of those who had dealt pense of building institutions.” This is a
the NATO forces in Afghanistan, the Tali- with police and reported sometimes hav- crucial point. Countries whose stability
ban have been fought to a stalemate. They ing to pay bribes is falling somewhat: from depends on an individual strongman are
seized a big city, Kunduz, in 2015, but were 53% in 2015 to 48% in 2016. brittle. Those that create inclusive institu-
driven out and have taken no more since. Foreign donors warm to Mr Ghani. In tions need never fail again. 7
The Economist January 7th 2017 49
Business
Also in this section
50 Toshiba’s latest write-down
51 Ford and Donald Trump
52 Schumpeter: The fat-cow years
2 the expense of the firm’s profits. In 2015 its valuable. It worked—its frozen-food sales consolidating procurement, which will
operating-profit margin was 15%, better in America grew faster. In November 2015 save about SFr2bn each year from 2020.
than the 13% at Danone, a French competi- they were 6% above what they had been a Whatever else Mr Schneider has on the
tor, but far below the 21% at Kraft-Heinz. year earlier. But Bernstein’s Andrew Wood menu for Nestlé, radical changes may be
Shareholders in the firm are waiting to see points out that the revival of frozen food somewhat limited by the fact that so many
whether Mr Schneider will shake things now looks wobbly again. of those who built the company into what
up. Some want him to sell off businesses Nor is Nestlé ignoring 3G’s strategy en- it is now are sticking around. Mr Bulcke is
that seem most at risk of long-term decline, tirely: it is trying to trim expenses. “We are expected to become its chairman. The out-
such as frozen food, as shoppers look for very much in an investment position, not going chairman, Peter Brabeck-Letmathe, a
fresher fare. in a cost-cutting exercise,” says Mr Roger, former Nestlé chief executive, may be-
“but that doesn’t mean that we don’t want come honorary chairman. Mr Bulcke, for
Food for life? to be cost-efficient in what we do.” One ef- one, seems sure that the company should
For now, Nestlé is defiant. “We started 150 fort, which includes trimming waste at fac- maintain its strong emphasis on the long
years ago having a product that actually— tories, is credited with saving about term. He taps his hand on the table, rattling
there’s symbolism there—saved the life ofa SFr1.5bn ($1.5bn) a year. Last year Nestlé an- some Nespresso cups, as he insists that
child,” says Paul Bulcke, the outgoing chief nounced organisational changes, such as growth is still the key. 7
executive. He and his colleagues say that
investment in health and related innova-
tion will produce strong growth at the Toshiba
company for years to come. Mr Schneider,
who used to run Fresenius, a big German
firm that offers kidney-dialysis products
Losing count
and other medical services, will certainly
emphasise that message. Nestlé differen-
tiates itself from 3G, with its keen focus on
cuts. Mr Roger says he respects what 3G
TOKYO
does, but that “they have a strategy which
Japan’s enfeebled giant faces a multi-billion-dollar write-down
is very different from ours.”
Still, few observers would call Nestlé a
health company. Many of its products are
perfectly healthy, including bottled water
T HE probe in 2015 into one of Japan’s
largest-ever accounting scandals, at
Toshiba, an electronics and nuclear-power
Westinghouse Electric, bought a nuclear-
construction firm, CB&I Stone & Webster.
One year on, on December 27th, Toshiba
and coffee. Many are not—milk chocolate conglomerate that has been the epitome of announced that cost overruns at that new
and ice cream, to name but two. And for the country’s engineering prowess, con- unit could lead to several billions of dollars
now, the purest forms of Nestlé’s focus on cluded that number-fiddling at the firm in charges against profits.
health contribute relatively little to its was “systemic”. It was found to have pad- Its shares fell by 42% in a three-day
sales. A business unit called Nestlé Health ded profits by ¥152bn ($1.3bn) between stretch as investors dumped them, fearing
Science, for example, sells nutritional pro- 2008 and 2014. Its boss, and half of the a write-down that could wipe out its share-
ducts for medical needs, such as vitamin- board’s 16 members, resigned; regulators holders’ equity, which in late September
packed drinks for the elderly and for can- imposed upon it a record fine of $60m. stood at $3.1bn. Moody’s and S&P, two rat-
cer patients. It contributes less than 5% of Now its deal-making nous is in doubt ings agencies, announced credit down-
revenue. too. In December 2015—the very same grades and threatened more. Toshiba’s ex-
The firm has a research institute de- month that it forecast hundreds of billions planation for how it got the numbers so
voted to studying food’s role in the man- of yen in losses for the financial year then wrong on a smallish purchase is woolly.
agement and prevention of disease—for ex- under way, as it struggled to recover from But it is clear that missing construction
ample, better understanding nutrition’s the scandal—Toshiba’s American arm, deadlines on nuclear-power plants can
ability to promote brain health. It may send costs skyrocketing. Its projects in
bring growth but probably only in the long America, and in China, are years behind
term. Nestlé has also partnered with schedule. Mycle Schneider, a nuclear ex-
young drugs firms, including one that is pert, says that in America, as elsewhere,
testing a treatment for ulcerative colitis. engineering problems are compounded by
More immediately rewarding may be a shortage of skilled manpower. Few
its efforts to make best-selling but un- plants have been built there recently.
healthy foods a bit more wholesome. In Part of the $229m that Westinghouse
November the company said it had created paid for CB&I Stone & Webster included
hollow sugar crystals that taste sweet but $87m of goodwill (a premium over the
contain fewer calories than the usual stuff. firm’s book value based on its physical as-
It will begin to put the new ingredient in its sets). It is that initial estimate that is now
chocolate in 2018. being recalculated.
It is also proud of changes to the mil- Toshiba had looked to be bouncing
lions of frozen dinners it sells every week back from its accounting nightmare. Before
in America. Shoppers had been avoiding the latest plunge it had made the second-
the frozen-food aisle. Nestlé first tried dis- biggest gains on the Nikkei 225 index in
counts, and then in 2015 introduced new 2016, where its shares were up by 77%. In
versions of its Lean Cuisine products, strip- April it wrote off $2.3bn on the goodwill
ping out unpalatable ingredients and re- value of Westinghouse, purchased for
placing them with organic ones. At $5.4bn in 2006—a write-down that it had
Stouffer’s, another frozen brand, Nestlé de- long avoided. In August it announced its
cided to target men with easy, protein- first profit in six quarters. It forecast a net
packed meals that are more nutritionally Ritual contrition profit of ¥145bn for the financial year of 1
The Economist January 7th 2017 Business 51
ACCOUNTING SCANDAL
CB&I STONE &
WEBSTER
WRITE-DOWN
Wheel spin
600
CEO RESIGNS
500 Ford Motors cancels a new plant in Mexico
RECORD LOSS $4.6BN
400
300
I T WAS in the spring of 2016 that Donald
Trump singled out Ford Motors, calling
its plans to build a plant in Mexico an
Rock facility, where Ford this week trum-
peted 700 new jobs to come, the firm had
already announced back in December
WESTINGHOUSE
200 “absolute disgrace” and promising it 2015 that it would invest in electrification
WRITE-DOWN
100 would not happen on his watch. Back and in 13 new electric vehicles. Linking
then, it seemed remarkable that the one location for that (Flat Rock) with the
0 candidate thought he could boss around Mexican plant cancellation looked like
2015 16 17
a firm of Ford’s stature. On January 3rd yet more accomplished spin.
Source: Thomson Reuters
Ford cancelled its $1.6bn project in the Things would undoubtedly be diffi-
Mexican state of San Luis Potosí and said cult for global carmakers if Mr Trump
2 2016-17, a clear reversal from its ¥460bn loss it would instead invest $700m into an tried to follow through on a campaign
of the previous year. Part of that was existing plant in Flat Rock, Michigan, to promise to slap a 35% tariff on cars ex-
thanks to a bold turnaround plan: firing build electric and autonomous cars. ported from Mexico to America. In 2015
14,000 staff, as well as selling lossmaking Ford’s manoeuvre seems more wheel- the country exported 2.7m vehicles, over
parts of its manufacturing empire, like TVs, spin than U-turn. Mr Trump’s strong- four-fifths of which went to North Ameri-
and one of its star units, a medical-equip- arming of corporate America is real ca. By appearing to kowtow to the new
ment maker, for $6bn. enough, and the carmaker will have boss-in-chief, Ford’s chief executive,
That left it free to focus on its semicon- gained much favour with the president- Mark Fields, may hope to keep this threat
ductor arm, which has been buoyed by de- elect. But its decision can be explained at bay—and to extract other favourable
mand from Chinese smartphone makers, largely in operational terms. The original concessions, such as softer rules on emis-
and its nuclear unit, which accounts for a plan was for the new Mexican plant to sions standards. “We have a president-
third of its revenue. The latest write-down build chiefly Focus cars—small passenger elect who has said very clearly that one
could dampen future investment in both. vehicles for which demand has fallen, of his first priorities is to grow the econ-
Toshiba has limited ways left to raise cash. thanks to America’s love affair with omy,” enthused Mr Fields. “That should
It has been barred from doing so on the SUVs, crossovers and pick-up trucks and be music to our ears.”
stockmarket ever since it was put on alert to low petrol prices. The decision to scrap Next in the line of fire is General Mo-
after the accounting fiasco—one step short the new plant looks far more like Ford tors, America’s biggest carmaker, which
of a delisting. reducing its exposure to the small-car said in 2013 that it would invest $5bn in
Observers reckon that Toshiba has game in North America than reducing its Mexico over six years. This week Mr
some room to manoeuvre, and that it will footprint in Mexico, says George Galliers Trump admonished it for making its
not ditch its nuclear business. It could raise at Evercore, an investment bank. Chevy Cruze, another compact car, most-
as much as $4bn from the sale of some The firm will still move production of ly over the border. “Make in U.S.A. or pay
part-owned subsidiaries, including Nu- the Focus away from its plant in Wayne, big border tax!” he tweeted. The com-
Flare, a spinoff of its semiconductor unit, Michigan to an existing plant in Hermosi- pany may find it hard to match Ford’s
says Seth Fischer of Oasis Management in llo, Mexico. As for the upgrade of the Flat skilful road-handling.
Hong Kong, a hedge fund, and a share-
holder in Toshiba’s power-station affiliate.
It could even choose to sell its lucrative
chip business altogether (Toshiba is the
world’s second-biggest maker of NAND
chips after Samsung Electronics of South
Korea), as well as some of its remaining
consumer-electronics ones.
Toshiba’s central part in a plan by the
government of Shinzo Abe, the prime min-
ister, to pep up growth by exporting nuc-
lear-power technology to emerging coun-
tries may help. In June Westinghouse
clinched a deal in India to build six new-
generation AP1000 reactors, Toshiba’s first
order since the triple meltdown at the Fu-
kushima Dai-ichi nuclear plant in 2011.
Toshiba is also involved in that site’s costly The consensus on Toshiba’s latest the scandal laid bare in 2015. Satoshi Tsuna-
and complex clean-up. Some think that screw-up is that a long-standing culture of kawa, who was installed as the company’s
Japanese banks, known for keeping zom- poor management is to blame. Toshiba’s new boss in June 2016, said last week that
bie firms on life support, will stand behind audit committee, for example, was until he had only become aware of the problem
it, come what may. Shares in Toshiba’s two 2015 headed by its former chieffinancial of- with CB&I Stone & Webster in December. It
main lenders, Sumitomo and Mizuho, slid ficer; such bodies should be fully indepen- was in 2015 that Mr Abe introduced Japan’s
last week after the profit warning. Inves- dent, says Nicholas Benes of the Board Di- first detailed rules on how companies
tors expect more big bank loans or a debt- rector Training Institute of Japan. It is not should run themselves. The spectacle of
for-equity swap, which allows a bank to clear whether or not the firm has fully over- Toshiba’s apparently endless crisis sug-
turn bad loans into shares. hauled its culture as part of its response to gests more needs to be done. 7
52 Business The Economist January 7th 2017
Banks in the rich world are getting their appetites back. Don’t be too scared
tigue seems to have set in among the public. True, when firms
misbehave, there is still a firestorm of outrage. John Stumpf, the
boss of Wells Fargo, quit in October after his bank admitted creat-
ing fake accounts. But many people can see that power has mi-
grated from banking to the technology elite in California. The
brew of high pay, monopolistic tendencies and huge profits that
attracts populist resentment is now more to be found in Silicon
Valley than in Wall Street or the City of London.
Global supervisors are still cooking up new rules, known as
“Basel 4” (see page 54), but are unlikely to demand a big rise in the
safety buffer the industry holds in the form of capital. The stron-
gest banks are signalling that they will lay out more in dividends
and buy-backs, rather than hoard even more capital (today, the
top 100 rich-world banks pay out about 40% of their profits).
A third reason for optimism in bank boardrooms is returns.
Global banking’s return on equity (ROE) has crept back towards a
respectable 10%. The worst of the fines imposed by American reg-
ulators are over. So far, “fintech” startups that use technology to
compete with rich-world banks have not won much market
share; banks have used technology to boost efficiency. They have
also got better at working out which of their activities create val-
Indian economics tors, such as the rise in the oil price and the
surge in the value of the dollar after the
Many rupee returns election of Donald Trump, are also at play.
Whether the costs of the exercise justify
the benefits depends, of course, on what
those benefits are. In his speech announc-
ing the measure, Narendra Modi, the prime
minister, highlighted combating corrup-
MUMBAI
tion and untaxed wealth. Gangsters and
The impact of India’s radical monetary reform is becoming clearer
profiteers with suitcases full of money
2 procedure requires trampling on the credi- vestment. Bodies such as the council of in- der management.
bility ofthe Reserve BankofIndia (RBI), the vestors and borrowers that sets the Green But that is to ignore the scale and pro-
central bank, which must first agree to dis- Bond Principles, guidelines for bonds ear- gress that large institutional investors have
honour the promise, on all banknotes, to marked for environmental projects, have brought to impact investing. Although
“pay the bearer” the value. If it does so, “ex- helped set common standards. $7bn is a tiny slice of Goldman’s portfolio,
tinguishing” the notes and its liability for Definitional squabbles still plague the it is huge compared with the investments
them, it can transfer an equivalent amount impact community. For sticklers, invest- of even well-established impact special-
to the government budget. ment only deserves “impact” status if it de- ists, such as LeapFrog, whose commit-
With so much cash handed in at banks, livers both near-market level returns and ments total around $1bn. And the entry of
the amount remitted to government by the strict measurement of the non-financial hard-nosed financial giants sends an im-
RBI might amount to perhaps 0.2-0.3% of impact: eg, of the carbon emissions saved portant message about impact investing:
GDP. Proceeds from a tax-amnesty scheme by a renewable-energy project; or of the that they see it as profitable for themselves
for cash-hoarders may swell the figure. number of poor people who borrow from and their clients. It is not enough to make
Even so, it will not be enough to justify the a microcredit institution. Others, however, investors feel good about themselves; they
costs of demonetisation—or even, perhaps, include philanthropic investment, where also want to make money. 7
the damage to the reputation of the RBI, financial returns are sacrificed for greater
which is already facing questions about its social benefits; or less rigorous types of do-
independence. But having imposed the good investments. Bank capital
costs, Mr Modi will be keen to trumpet Such disagreements make it hard to
whatever benefits he can find. 7 gauge the true extent of impact invest-
ment. For instance, BlackRock Impact and
Polishing the floor
Goldman both also offer two looser invest-
Impact investing ment categories: “negative screening” (ie,
not investing in “bad” sectors—say, tobacco
Coming of age or oil); and “integrated” investments that
take environmental, social or governance
Supervisors put off finalising reforms to
the Basel rules
(ESG) considerations into account (eg, by
selecting for firms with, say, good working
conditions). Neither firm, however, pro-
vides a complete breakdown of these cate-
S OME banks find existing capital require-
ments too taxing. To no one’s surprise,
on December 23rd Monte dei Paschi di Si-
Investing to do good as well as to make
gories by assets under management. ena, at present Italy’s fourth-biggest bank,
money is catching on
The industry is also held back by a re- asked the Italian state for help, having
2 churned out by a “standardised” method. ed; several European lenders could be means fixing a floor, but how high? Omar
The higher the percentage, the tighter the stung. That is partly because America has Keenan and Kinner Lakhani, of Deutsche
standard: a first version of the proposals already installed floors in its domestic Bank, estimate that a 75% floor would in-
suggested 60-90%; a failed compromise rules—and, Americans would add, its crease the RWAs (and hence reduce the
last month proposed gradually raising it to banks shaped up faster after the crisis. capital ratios) of 26 of the 34 listed Euro-
75% over four years, starting in 2021. Europeans retort that it also reflects trans- pean banks they cover; at 60%, the number
American officials like the floor, believ- atlantic differences in business models. drops to ten, mainly in the Netherlands
ing that it limits banks’ ability to play European lenders tend to keep more resi- and Nordic countries.
games with the rules. European banks and dential mortgages on their books than Phasing in the rules would give banks
officials don’t. Both the Association of Ger- American banks, which often sell them on; time to adapt. Under the timetable envis-
man Banks and the Bundesbank, for exam- they also lend more to companies and for aged by the committee, they would have
ple, want no floor at all. They argue that in- project finance. All this may carry heavier until 2025—almost two decades after the
ternal models make capital calculations risk-weights under the revised rules. world’s financial system started to crack. If
more, not less, sensitive to risk. Officials are still aiming for agreement the stand-off continues, the repairs will
America’s banks would be little affect- in the first quarter of 2017. That probably take even longer to complete. 7
The world is changing and investors may be too optimistic about the results
Anthony Atkinson
Insuring talent
For poorer, for Death Star
richer
When the famous die, it is increasingly costly for insurers
2 erative disease that he works on. Both mol- from this information the network of pro- drug or not. Most known protein struc-
ecules remain confidential while their util- tein interactions that underlie that disease. tures have been worked out from crystal-
ity is being assessed. One is bang in the At that point human researchers intervene lised versions of the molecule, held tight
middle of what he and his team are doing to test the model’s predictions in a real bio- by networks of chemical bonds. In reality,
already. To him, this confirms that the artifi- logical system. One of the potential drugs proteins are flexible, but that is much hard-
cial intelligence in question is generating BERG Health has discovered this way—for er to deal with.
good ideas. The other, though, is compli- topical squamous-cell carcinoma, a form More work at the molecular level is
cated and not obvious, but mechanistical- of skin cancer—passed early trials for safe- therefore needed before AI will be able to
ly interesting. Without the AI to prompt ty and efficacy, and now awaits full-scale crack open the inner workings of a cell.
them, it is something his team might have testing. The company says it has others in One of CZI’s first projects is generating just
ignored—and that, he admits, might in turn development. such basic data. That, in itself, is a massive
be a result of their bias. For all the grand aspirations of the AI undertaking—but it is one which collabora-
For now, BenevolentAI is a small actor folk, though, there are reasons for caution. tion with artificial intelligence will also
in the theatre of biology and artificial intel- Dr Mead warns: “I don’t think we are in a speed up. AI will nudge people to generate
ligence. But much larger firms are also in- state to model even a single cell. The model new data and run particular experiments.
volved. Watson, a computer system built we have is incomplete.” Actually, that in- Those people will then ask the AI to sift the
by IBM, is being applied in similar ways. In completeness applies even to models of results and make connections. As Isaac
particular, IBM has gone into partnership single proteins, meaning that science is not Newton put it, “If I have seen further, it is
with Pfizer, an American pharma com- yet good at predicting whether a particular by standing on the shoulders of giants.” If
pany, with the intention of accelerating modification will make a molecule intend- the brains of those giants happen to be
drug discovery in immuno-oncology—a ed to interact with a given protein a better made of silicon chips, so be it. 7
promising area of cancer therapy that en-
courages the body’s own immune system
Olfactory medicine
to fight tumours.
Artificial intelligence will also move
into clinical care. Antonio Criminsi, who, Whiff of danger
like Dr Bishop, works at Microsoft Research
in Cambridge, observes that today the pro-
A prototype device to detect the smell of disease
cess of delineating the edges of tumours in
images generated by MRI machines and CT
scans is done by hand. This is tedious and
long-winded (it can take up to four hours).
O NE of a doctor’s most valuable tools
is his nose. Since ancient times,
medics have relied on their sense of
able manner. The combined changes
generated an electrical fingerprint that,
the researchers hoped, would be diag-
AI can reduce the time taken to minutes, or smell to help them work out what is nostic of the disease a patient was suf-
even seconds—and the results are com- wrong with their patients. Fruity odours fering from.
pletely consistent, unlike those arrived at on the breath, for example, let them To test their invention, Dr Haick and
by human doctors. monitor the condition of diabetics. Foul his colleagues collected 2,808 breath
Another example of AI’s move into the ones assist the diagnosis of respiratory- samples from 1,404 patients who were
clinic is described in a recent paper in JAMA, tract infections. suffering from at least one of the diseases
an American medical journal. This paper But doctors can, as it were, smell only they were looking at. Its success varied. It
showed that it is possible to use AI to detect what they can smell—and many com- could distinguish between samples from
diabetic retinopathy and macular oedema, pounds characteristic of disease are patients suffering from gastric cancer and
two causes of blindness, in pictures of the odourless. To deal with this limitation bladder cancer only 64% of the time. At
retina. Enlitic, a new firm based in San Hossam Haick, a chemical engineer at distinguishing lung cancer from head and
Francisco, is using AI to make commercial the Technion Israel Institute of Tech- neck cancer it was, though, 100% success-
software that can assist clinical decisions, nology, in Haifa, has developed a device ful. Overall, it got things right 86% of the
including a system that will screen chest X- which, he claims, can do work that the time. Not perfect, then, but a useful aid to
rays for signs of disease. Your.MD, a firm human nose cannot. a doctor planning to conduct further
based in London, is using AI, via an app, to The idea behind Dr Haick’s invention investigations. And this is only a proto-
offer diagnoses based on patients’ queries is not new. Many diagnostic “breathalys- type. Tweaked, its success rate would be
about symptoms. IBM is also, via Watson, ers” already exist, and sniffer dogs, too, expected to improve.
involved in clinical work. It is able to sug- can be trained to detect illnesses such as
gest treatment plans for a number of differ- cancer. Most of these approaches,
ent cancers. All this has the potential to though, are disease-specific. Dr Haick
transform doctors’ abilities to screen for wanted to generalise the process.
and diagnose disease. As he describes in ACS Nano, he and
his colleagues created an array of elec-
The power of networking trodes made of carbon nanotubes (hol-
Another important biological hurdle that low, cylindrical sheets of carbon atoms)
AI can help people surmount is complex- and tiny particles of gold. Each of these
ity. Experimental science progresses by had one of 20 organic films laid over it.
holding steady one variable at a time, an Each film was sensitive to one of a score
approach that is not always easy when of compounds known to be found on the
dealing with networks of genes, proteins breath of patients suffering from a range
or other molecules. AI can handle this of17 illnesses, including Parkinson’s
more easily than human beings. disease, multiple sclerosis, bladder can-
At BERG Health, the firm’s AI system cer, pulmonary hypertension and
starts by analysing tissue samples, geno- Crohn’s disease. When a film reacted, its
mics and other clinical data relevant to a electrical resistance changed in a predict- The nose knows
particular disease. It then tries to model
62 Science and technology The Economist January 7th 2017
Atmospheric physics
Palaeontology
The storm before Cracking a puzzle
the calm
How reptilian were dinosaur eggs?
Britain and the European Union draft ofhistory. It will not give Mr Cameron
much satisfaction.
Why Brexit won Partly because they expected to win
easily, as Harold Wilson did in 1975, Mr
Cameron and the Remainers made tactical
mistakes. These included accepting a pre-
vote period of official government “pur-
dah”, constraining what it could publish;
allowing cabinet ministers to back Leave
The first crop of Brexit books includes entries rich with detail and analysis
without resigning; and avoiding direct
2 failed to answer the economic argument, endum harder. Although he nominally many pollsters got the result wrong.
being reduced to Mr Gove’s notorious at- backed Remain, he and his team often sab- The third goes back to Mr Hannan and
tack on “experts”. They did not set out clear otaged the Labour In campaign, for exam- his friends. For three decades British gov-
alternatives to membership. Their internal ple refusing to use the word “united” to de- ernments of both parties, egged on by a
splits and focus on immigration often scribe Labour’s position or to share a shrilly Eurosceptic press, did little but carp
made them seem nasty, a big worry when platform with former party leaders. at Brussels. Mr Cameron’s delusion was
a Labour MP, Jo Cox, was brutally mur- A second was the rising anti-elite, anti- that, having himself hinted that he might
dered in mid-June, just before the vote, by a London and anti-globalisation mood of campaign to Leave, he could turn senti-
man linked to the far right. By then many many voters, especially in the Midlands ment round completely in just three
Leavers thought they would lose. and north. Those who feel they were left months. Instead, his past stance made him
That they won is down to three causes behind after the financial crisis have seem unconvincing when he portrayed EU
deeper than Remainers’ tactical errors. turned to populists in many countries (in- membership as vital for Britain’s economy
One was the Labour Party leadership. The cluding to Donald Trump in America). In and security. This same legacy could now
arrival of Jeremy Corbyn, a far-left anti-EU the Brexit referendum they voted in unex- make it trickier for Mrs May to persuade
figure, in late 2015 made winning the refer- pectedly large numbers, a big reason why voters to accept a soft Brexit. 7
The past12 months saw many words enjoy a breakthrough. Unfortunately most of them are grim
Chinese economics
Fiction
Western takeaway Crazy city
2 book sold more than 100,000 copies. pening on factory floors or in farm fields. a Basel firm that also specialises in muse-
The World Bank also had a big hand in But it is still a gripping read, highlighting ums, completed 1111Lincoln Road in 2010. A
China’s take-off. The bank has a tainted what was little short of a revolution in Chi- ziggurat of bare concrete linked by precipi-
reputation from that era, when it was seen na’s economic thought. tous ramps, it provides accommodation
as pushing a “Washington consensus” Reading the book today, it is tempting to for a series of art-crowd-friendly shops on
agenda of liberalisation that harmed Latin conclude that China is ignoring a basic les- the ground floor and a home for the devel-
America. Much less attention is paid to its son from its success: that being open to for- oper, Robert Wennett, on the roof. This gid-
subtler positions in China in the 1980s. It eign ideas served it so well. Under Xi Jin- dy stack of concrete cards set a benchmark
carried out two major studies of the econ- ping, officials rail against “Western values”. for audacity, its upper deck providing stun-
omy (the first of their kind), became Chi- Yet there is also a less gloomy conclusion. ning views and one of the most sought-
na’s largest source of foreign capital and, China’s path has never been linear: re- after party spaces during Art Basel Miami
responding to Chinese requests, provided formists and conservatives have constant- Beach. From this example, the high-end car
reams of useful policy advice. ly jostled for the upper hand. But voices for park became firmly established.
Mr Gewirtz’s book does not attempt to openness have ultimately prevailed. And In November, as part of a new six-block
provide a definitive account of China’s the gains that China has made in its under- development in the mid-Beach area, Alan
economic rise. It dwells in the world of standing of economics and, more funda- Faena, an Argentine developer, revealed
ideas, tracing the arc of debates. Little at- mentally, in the lives of its people will not his parking garage (pictured). It boasts a
tention is paid to what was actually hap- be easily undone. 7 glazed side elevation that exposes the ro-
botic car elevator, which installs and re-
trieves cars from closely stacked shelves: a
Architecture preparation for the dance performances in
the Faena Forum arts centre to which it is
Pile ‘em in style appended. The car park actually only pro-
vides room for around 100 cars (though
there are 300 subterranean parking places
beneath the development). Yet still Mr Fae-
na felt that the development needed an
above-ground car park, to be “a state-
ment”. He had his designed, like the adja-
The most exciting architecture in Miami Beach is car parks
cent arts centre, by OMA, the fashionable
Tenders
Markets
% change on GDP forecasts
Dec 31st 2015 2017, % change on a year earlier
Index one in local in $ Best Worst
Jan 4th week currency terms
0 2 4 6 8 10 6 4 2 – 0 + 2
United States (DJIA) 19,942.2 +0.5 +14.4 +14.4
China (SSEA) 3,307.5 +1.8 -10.7 -16.6 Myanmar Trinidad & Tobago
Japan (Nikkei 225) 19,594.2 +1.0 +2.9 +5.6
Ivory Coast Ecuador
Britain (FTSE 100) 7,189.7 +1.2 +15.2 -3.9
Canada (S&P TSX) 15,516.8 +1.0 +19.3 +24.6 Bhutan Azerbaijan
Euro area (FTSE Euro 100) 1,121.9 +1.1 +2.5 -1.1
Euro area (EURO STOXX 50) 3,317.5 +1.2 +1.5 -2.1 Laos Chad
Austria (ATX) 2,682.6 +1.7 +11.9 +7.9 Cambodia Syria
Belgium (Bel 20) 3,665.7 +1.6 -0.9 -4.5
France (CAC 40) 4,899.4 +1.1 +5.7 +1.9 Tanzania Timor-Leste
Germany (DAX)* 11,584.3 +1.0 +7.8 +4.0
Ghana Puerto Rico
Greece (Athex Comp) 657.5 +3.4 +4.1 +0.4
Italy (FTSE/MIB) 19,626.6 +2.0 -8.4 -11.6 India Equatorial Guinea
Netherlands (AEX) 487.6 +0.7 +10.4 +6.4
Djibouti Libya
Spain (Madrid SE) 956.1 +1.5 -0.9 -4.5
Czech Republic (PX) 934.2 +1.2 -2.3 -5.8 Vietnam Venezuela
Denmark (OMXCB) 805.4 +1.0 -11.2 -14.0
Source: Economist Intelligence Unit
Hungary (BUX) 32,649.0 +1.9 +36.5 +35.0
Norway (OSEAX) 772.6 +0.8 +19.1 +22.5
Poland (WIG) 52,753.8 +2.8 +13.5 +7.5 Other markets The Economist commodity-price index
Russia (RTS, $ terms) 1,176.7 +3.4 +28.9 +55.4 % change on 2005=100
% change on
Sweden (OMXS30) 1,530.9 +0.2 +5.8 -2.0 Dec 31st 2015 Dec 20th Dec 27th Jan 3rd one one
Switzerland (SMI) 8,354.8 +1.2 -5.3 -7.4 Index 2016 2016 2017* month year
one in local in $
Turkey (BIST) 76,143.6 -1.8 +6.2 -13.3 Jan 4th week currency terms Dollar index
Australia (All Ord.) 5,788.2 +1.0 +8.3 +8.4 United States (S&P 500) 2,270.8 +0.9 +11.1 +11.1 All items 142.0 140.8 141.9 -1.7 +13.8
Hong Kong (Hang Seng) 22,134.5 +1.7 +1.0 +0.9 United States (NAScomp) 5,477.0 +0.7 +9.4 +9.4
India (BSE) 26,633.1 +1.6 +2.0 -0.9 Food 154.9 152.6 154.8 -1.2 +6.4
China (SSEB, $ terms) 344.8 +1.1 -13.5 -19.1
Indonesia (JSX) 5,301.2 +1.8 +15.4 +18.4 Japan (Topix) 1,554.5 +1.2 +0.5 +3.1 Industrials
Malaysia (KLSE) 1,647.5 +1.1 -2.7 -7.1 Europe (FTSEurofirst 300) 1,443.8 +1.0 +0.4 -3.1 All 128.6 128.5 128.5 -2.3 +24.8
Pakistan (KSE) 48,705.0 +2.7 +48.4 +48.3 World, dev'd (MSCI) 1,774.0 +1.3 +6.7 +6.7 Nfa† 136.5 136.8 138.1 +2.4 +27.9
Singapore (STI) 2,921.3 +0.8 +1.3 -0.2 Emerging markets (MSCI) 871.5 +2.4 +9.7 +9.7 Metals 125.2 124.9 124.4 -4.3 +23.4
South Korea (KOSPI) 2,045.6 +1.0 +4.3 +1.4 World, all (MSCI) 427.2 +1.4 +7.0 +7.0 Sterling index
Taiwan (TWI) 9,287.0 +0.9 +11.4 +13.5 World bonds (Citigroup) 877.6 +0.2 +0.9 +0.9
Thailand (SET) 1,563.6 +2.6 +21.4 +22.0 All items 209.0 208.9 210.8 +2.1 +36.2
EMBI+ (JPMorgan) 776.0 +0.7 +10.2 +10.2
Argentina (MERV) 18,143.1 +9.9 +55.4 +24.9 Hedge funds (HFRX) 1,203.2§ -0.1 +2.5 +2.5 Euro index
Brazil (BVSP) 61,589.1 +3.0 +42.1 +74.4 Volatility, US (VIX) 11.9 +13.0 +18.2 (levels) All items 170.1 167.4 169.9 +1.6 +17.5
Chile (IGPA) 20,809.5 +1.4 +14.6 +20.9 CDSs, Eur (iTRAXX)† 67.9 -5.5 -12.0 -15.1 Gold
Colombia (IGBC) 10,288.4 +1.6 +20.4 +28.9 CDSs, N Am (CDX)† 63.4 -6.6 -28.2 -28.2 $ per oz 1,131.6 1,133.5 1,156.1 -1.3 +7.3
Mexico (IPC) 46,587.7 +2.2 +8.4 -12.5 Carbon trading (EU ETS) € 5.7 -10.0 -31.6 -34.1 West Texas Intermediate
Venezuela (IBC) 31,839.2 +4.7 +118 na Sources: Markit; Thomson Reuters. *Total return index.
Egypt (EGX 30) 12,608.4 +2.8 +80.0 -23.3 †Credit-default-swap spreads, basis points. §Dec 29th. $ per barrel 51.9 52.0 52.3 +2.7 +45.9
Sources: Bloomberg; CME Group; Cotlook; Darmenn & Curl; FT; ICCO;
Israel (TA-100) 1,287.8 -0.1 -2.1 -1.1
Indicators for more countries and additional ICO; ISO; Live Rice Index; LME; NZ Wool Services; Thompson Lloyd &
Saudi Arabia (Tadawul) 7,198.1 -0.6 +4.1 +4.2 Ewart; Thomson Reuters; Urner Barry; WSJ. *Provisional
South Africa (JSE AS) 50,760.2 +0.9 +0.1 +14.1 series, go to: Economist.com/indicators †Non-food agriculturals.
70 The Economist January 7th 2017
Obituary Vera Rubin
their geometry in order to look them up lat-
er. He even helped her make her first tele-
scope, from a cardboard tube; she had al-
ready made her own kaleidoscope. She
hadn’t ever met an astronomer, but it never
occurred to her that she couldn’t be one.
But her early research was largely ig-
nored. In other work, male astronomers el-
bowed her aside. Fed up, she looked for a
problem “that people would be interested
in, but not so interested in that anyone
would bother me before I was done.”
She found it. In the 1930s Fritz Zwicky,
an idiosyncratic Swiss astrophysicist, had
suggested that the brightly shining stars
represented only a part of the cosmic
whole. There must also be “dark matter”,
unseen but revealed indirectly by the ef-
fects of its gravity. That conjecture lan-
guished on the margins until Ms Rubin,
working with her colleague Kent Ford, ex-
amined the puzzle of galactic rotation. Spi-
ral galaxies such as Andromeda, she
proved, were spinning so fast that their
outer stars should be flying away into the
never-never. They weren’t. So either Ein-
stein was wrong about gravity, or gravita-
tional pull from vast amounts of some-
Dark star thing invisible—dark matter—was holding
the stars together.
The discovery reshaped cosmology,
though initially her colleagues embraced it
unenthusiastically. Astronomers had
thought they were studying the whole uni-
Vera Rubin, an American astronomer who established the existence of dark matter,
verse, not just a small luminous fraction of
died on December 25th, aged 88
it. New theories developed on what the
SUBSCRIBERS ENJOY:
Espresso
N 9 05
Yet to subscribe? Visit Economist.com/newyear to get started for just $12 for 12 weeks
and enjoy access across print, online, audio and via our apps.
Slack is where work happens, for millions
of people around the world, every day.