Tuesday 3 March 2020 ★ † FINANCIAL TIMES 11

Opinion

A

friend of Frank Sinatra

used to tell a joke about Ol’

Blue Eyes. “You know

Frank saved my life once.

These five guys were beat-

ing me up and Frank said, ‘that’s enough

boys’.” You could make the same crack

about Boris Johnson’s 10 Downing Street

operation, whose guys have been beat-

ing up a lot of people, although the UK

prime minister has yet to say “that’s

enough”.

All leaders need warriors, but this

administration is led by people temper-

amentally drawn to conflict. “They are

campaigners. That’s who they are,” says

one ally. But from battles with White-

hall, judges and the media to feuding

within, this mindset is creating prob-

lems a government already facing the

challenges of Brexit does not need.

In all its feuds, Downing Street has a

point and sometimes a strong one.

There is nothing unprecedented in a

minister wanting to remove her top civil

servant, even though Philip Rutnam’s

departure as the top Home Office offi-

cial over the weekend was more spec-

tacular than normal. A prime minister

is entitled to seek Whitehall reform and

wise to want to end friction with the

Treasury. The government is right that

technology raises questions about the

BBC’s future funding. There is also an

issue around the scope of judicial

reviews (although Mr Johnson seemed

able to forgive the one last week that

delivered the coup de grâce to Heath-

row’s expansion).

But the disproportionate rhetoric and

response — from losing a chancellor to

pledging to “whack” the BBC — suggest

an unconservative readiness to under-

mine any institution which might offer a

check on its ambitions. It also points to a

real fragility. The impression is of a lead-

ership happiest in the bunker sniffing

out enemies and threats to its power.

Through petulance, perturbation and

pugnacity, this begins to look like an

administration more comfortable with

campaigning than governing.

Contrast this with Mr Johnson’s slug-

gish early approach to coronavirus. The

health secretary, Matt Hancock, was left

to manage the first stages while ministe-

rial meetings were conducted at sub-

cabinet level. It was only at the weekend

that Mr Johnson decided to step in. It is

hard not to conclude that the issue did

not excite either him or those around

him. There is no obvious enemy, bar

germs, and no three-word slogan to rally

round — except “wash your hands”.

At one level the Rutnam case is sui

generis. Events like his falling out with

Priti Patel, the home secretary, happen.

Those who worked with Sir Philip know

his flaws. “He was no angel,” said one.

He presided over a dysfunctional

department of fiefdoms and fell out

with two home secretaries. But Ms Patel

will not come out of this well. No doubt

institutional obduracy was slowing

immigration reform but she arrived

with an outsider’s temperament, impa-

tience with complexity and a Nike slo-

gan approach to management. Downing

Street knows her shortcomings which is

why it keeps sending trusted aides to

manage her. But she is an ally and that

matters more.

What takes this beyond the specific is

that Sir Philip faced a hail of hostile

briefing which has now backfired into a

potential lawsuit. Officials at the Treas-

ury and Foreign Office were also warned

via media that they are on a blacklist.

Again, it is not unusual for an incom-

ing government to want to ginger up the

civil service. Both Margaret Thatcher

and Tony Blair did so. Dominic Cum-

mings and Michael Gove, the prime

minister’s chief strategist and cabinet

fixer, share Mrs Thatcher’s view of the

civil service as a thwarter of progress.

But it is unheard of for the heads of all

three major departments to be targeted.

It is also a mistake. The last thing that a

government this radical needs is White-

hall ducking into a defensive crouch and

burying frank advice.

The counterargument runs that the

Johnsonites are heeding Mr Blair’s

lament that it took him too long to make

changes and so are moving fast while at

the peak of their power. Yet Mr Blair’s

key to progress was the delivery unit

that he set up with its detailed focus on

targets and accountability. The irony is

that Mr Cummings, a believer in proc-

ess, knows this.

The coronavirus, with its potential to

derail the government’s agenda, may

mark a moment of emotional transition

from campaigning to governing, though

the paraphernalia of war rooms plays to

aggressive tendencies. But tackling it is

about detail, process and difficult deci-

sions. Mr Johnson is about to find out

how much he needs some of his

appointed enemies and the danger of

wars on multiple fronts. Is it too much to

hope Ol’ Blond Top will tell his aides it’s

time to stop the beatings?

[email protected]

The impression is of a

leadership happiest in the

bunker sniffing out

enemies and threats

Coronavirus means Johnson must govern not campaign

the EU will decide on breaches unilater-

ally. Having insisted on British sover-

eignty, they can hardly complain if the

EU side also makes sovereign decisions.

As for Brussels, it will have to accept

that a review process will mean that

there is an element of constant negotia-

tion in the UK-EU relationship. This is a

feature of the EU-Switzerland relation-

ship that Brussels is desperate to avoid

replicating.

There is also clearly a risk that the

British would game the system, by mak-

ing incremental changes that eventually

amount to significant divergence.

Even so, both sides will make real

gains from the sword of Damocles solu-

tion. The EU avoids the danger of a total

regulatory free-for-all. Britain retains

a decent level of access to its most

important market. And both sides

retain a co-operative relationship with

like-minded neighbours in an increas-

ingly dangerous world.

It is less exciting than a first-round

knockout. But it makes more sense.

[email protected]

practice opens the way for an eventual

deal. The British will insist on their right

to diverge from EU law but will assure

the Europeans that, in practice, they

will retain social and environmental

standards that are at least as stringent as

those of the EU.

This is rather like a child insisting that

it will determine its own bedtime but

indicating that, in practice, it will go to

bed at 8pm as normal.

The EU, naturally, will not take this

pledge on trust. It will grant Britain tar-

iff and quota-free access to the Euro-

pean single market. But it will also set

up a formal process through which the

EU reviews British legislation.

If the European Commission or par-

liament (or some combination) finds

that the UK has un-levelled the playing

field, the EU will retain the right to

impose tariffs on Britain.

Any such deal would involve risks and

compromises for both sides. The British

will have to trade with a sword of Damo-

cles over their heads, which might dis-

courage investment.

The UK will also have to accept that

Mr Johnson’s Conservative party are los-

ing the ideological battle. They argued

for 30 years that Brussels regulation was

strangling the British economy and that

the UK should break free to opt for a

low-tax, small-state model.

But the pro-Brexit voters in the north

of England, who delivered the last elec-

tion for Mr Johnson, have been prom-

ised higher public spending and more

social protection. The idea of Britain as

“Singapore-on-Thames”, always a mis-

leading analogy, is sliding off the

agenda.

What remains of the original Brexit

agenda is the Johnson government’s

determination to restore British sover-

eignty, removing the EU’s right to legis-

late directly or indirectly for the UK.

This distinction between theory and

rapidly.” Others take a similar view.

There is no doubt that the issues

involved are complex and the timetable

is tight. The British side is belligerent

and unprincipled. The EU side is com-

placent and inflexible.

None of that looks good. But, despite

all this, I think a trade deal will be agreed

by the end of the year. I even have a fair

idea what it will look like.

Why the confidence? Partly, because

we saw both the EU and Britain flirt with

no deal during the negotiation of the

withdrawal agreement last year, and

then back off. Ultimately, neither side

was prepared to accept the destructive

consequences. It is certainly possible

that one or both sides will miscalculate

this time and we will end up with no

deal. But the odds are that, as with the

withdrawal agreement, a compromise

will be found.

As Eurasia’s Mr Rahman has

observed, the biggest issue in the negoti-

ations is Britain’s “super hardline on

divergence and the need to square this

with EU’s expectations on the level play-

ing field”. For those unfamiliar with the

jargon, this means that the UK insists

that it must retain the right to diverge

from EU law. In response, the EU insists

that if Britain undercuts EU standards

on labour law, the environment or state

subsidies, then it cannot also expect tar-

iff-free and quota-free access to the EU’s

single market.

It sounds like a big problem. Fortu-

nately, the argument may be more

about theory than reality. That is

because the Thatcherite purists within

T

he prelude to the trade

negotiations between Brit-

ain and the EU has felt like

the weigh-in for a heavy-

weight title fight.

As they prepared for Brexit Two, the

rematch, the two boxers shouted

threats at each other. Boris Johnson, the

Blond Bomber — fresh from a knockout

victory in the UK election — swore that

he would come out swinging. Britain, he

said, was ready to break off talks, rather

than take a step backwards and com-

promise. Boris’s corner seemed confi-

dent that Michel Barnier — the Bruiser

from Brussels (still undefeated, after

two years of brutal negotiations) — was

now ripe for the taking.

Meanwhile Mr Barnier’s trainer, Jean-

Yves Le Drian (also known as the

French foreign minister) warns that

Britain and the EU will “rip each other

apart”. On Monday, March 2, the bell

rang for the start of the discussions, and

the two boxers headed out into the ring.

It is all very exciting and very worry-

ing, even for experienced observers.

Mujtaba Rahman, head of the Eurasia

Group in Europe, warned last week

that: “Odds of no deal are rising and

space for a deal is shrinking — and

Brexit solved

with the sword

of Damocles

Odds are that, as

with the withdrawal

agreement, a compromise

will be found

I

t is a truism of our age that we suffer

from a deficit of trust. But in some

areas of technology, the opposite

is true: there is an alarming surfeit

of trust.

That was one of the findings of the

National Transportation Safety Board’s

report published last week into the

fatal crash of a semi-automated Tesla

in California in 2018. The investigators

concluded that the Model X sport utility

vehicle failed to read the road cond-

itions correctly and accelerated into

a crash barrier at 70.8mph while the

38-year-old driver was playing a game

on his iPhone.

As Robert Sumwalt, NTSB’s chair-

man, said, the crash was not only caused

by flaws in Tesla’s autopilot. It also

resulted from an over-reliance on

technology and human distraction.

“The lessons from this investigation are

as much about people as they are about

the limitations of emerging technol-

ogies,” he said.

“There is not a vehicle currently

available to US consumers that is self-

driving. Period. Every vehicle sold to US

consumers still requires the driver to be

actively engaged in the driving task,

even when advanced driver assistance

systems are activated.”

The tendency for humans to overtrust

automated decision-making systems

has a long and tragic history. Software

bugs have caused all manner of societal

malfunctions. Millions of people have

lost money on investments, received

incorrect medical diagnoses, been

falsely arrested, given excessive prison

sentences and died in aeroplane crashes

because of software failures. Yet all

too often we follow faulty automated

instructions because “the computer

can’t be wrong”.

The problem is growing more acute as

we entrust ever more decisions to ever

more complex computers. Some com-

puter scientists warn we risk relying on

artificial intelligence systems for con-

sistency and explainability rather than

acknowledging uncertainty and accept-

ing variance of outcomes.

Neil Lawrence, professor of machine

learning at Cambridge university, says

he is increasingly nervous about the

unrealistic promises made by some in

the AI industry that they can inject com-

mon sense into computer systems and

infer intent. “Up until now, every tech-

nology has been adapted to our needs.

But we say we are now building the first

systems that are responsive and adap-

tive to our needs,” he says.

The trouble is that human prefer-

ences are often incoherent, incomplete,

inconsistent or mistaken. Uncertainty

is endemic and circumstances change.

“The physical world is constantly

kicking you. The real world does strange

things,” says Prof Lawrence.

That is not to deny that computer

models can be incredibly useful, even in

conditions of uncertainty. But their lim-

itations must be constantly kept in

mind. In a paper, Decision Making and

Diversity, Prof Lawrence quotes the

statistician George Box: “Since all mod-

els are wrong the scientist must be alert

to what is importantly wrong. It is inap-

propriate to be concerned about mice

when there are tigers abroad.”

How we collectively learn from the

flaws of machines may be just as impor-

tant as understanding the failings of

humans. In his bookWho’s Driving Inno-

vation?, Jack Stilgoe argues that

advances in machine learning must be

accompanied by social learning.

Two general regulatory models exist,

but perhaps more need to be invented.

The first is “anticipatory governance”,

as practised by the US Food and Drug

Administration, which regulates the

pharmaceutical industry. Drugs are

subject to extensive clinical trials and

regulatory review before they are

released to the public. The second is

“governance by accident”, as followed

by the NTSB, which sounds alarming

but has proved astonishingly effective in

improving airline safety.

“Either you ‘protect and provide’ or

‘wait and see’,” says Mr Stilgoe, a social

science lecturer at University College

London. But what is essential, he sug-

gests, is to create a collective societal

capacity to understand emerging tech-

nologies and decide on the appropriate

regulatory framework. We cannot leave

all this to powerful private corporations.

Companies certainly have a duty to

refrain from launching inadequately

tested products on to the market. But

every user has a responsibility to exer-

cise due caution, too.

The US National Park Service urges

all visitors to Death Valley to carry

up-to-date maps and exercise common

sense because of unreliable connectivity

in the area. “Do not depend only on

your vehicle GPS navigation system,”

it warns.

That is a general warning we should

all observe.

[email protected]

We follow faulty

automated instructions

because ‘the computer

can’t be wrong’

Trusting AI too much can be fatal

TECHNOLOGY

John

Thornhill

BRITAIN

Robert

Shrimsley

global affairs

Gideon

Rachman

I

n 2016, Donald Trump won the US

presidency by promising to reverse

the trends of the past few decades:

the rich had got richer and the little

guy felt left behind. Four years

later, inequality is still on the agenda.

The working class is doing better than

it was a few years ago, with unemploy-

ment near record lows and wages rising.

Even so, the rich have pulled further

ahead. There are as many ways to meas-

ure US income inequality as there are

Trump tweets in a given week, but it

remains clear that it is still on a decades-

long march upwards.

In last month’s State of the Union

address, Mr Trump boasted of an all-

time high in median household

incomes. Median incomes do typically

rise during a recovery, but they don’t tell

you what’s happening at different

income levels. More granular data show

that the rich are still winning: incomes

for the top 10 per cent continued to rise

relative to the median.

Low-income wage growth has out-

paced that of middle- and high- wage

earners in the last year. That is likely to

be a function of the US Federal Reserve’s

efforts to keep the labour market hot in

order to pull marginal workers in from

the sidelines. Minimum wage laws in

dozens of US states and cities are also

playing a role.

But those gains will probably abate

when the labour market cools and mini-

mum wages level out. And wages are

only one form of income. The US Census

Bureau’s Current Population Survey

defines more than 20 other types of

income, including capital gains.

Economists disagree about the degree

to which inequality is rising. The Con-

gressional Budget Office estimates how

much of US income is claimed by the top

1 per cent of the population. Its surveys

show an upward trend, from 7.5 per cent

in 1979 to 12.5 per cent in 2016 after

taxes and transfers. The CBO forecasts

a further rise to 13.4 per cent in 2021.

Economists Thomas Piketty,

Emmanuel Saez and Gabriel Zucman

think inequality is worse than that.

Using tax data, they estimate that the

top 1 per cent’s share of income rose

from 9.1 per cent in 1979 to 15.7 per cent

in 2014 after taxes. Using different cal-

culations, economists Gerald Auten and

David Splinter insist the gains of the 1

per cent are much more modest, with

their share of income rising from 7.2 per

cent in 1979 to 8.6 per cent in 2014 after

taxes and transfers.

Whatever their precise share, the rich

have been getting richer than the rest

for decades. A common measure of

income inequality is the Gini coefficient

— a value of zero reflects perfect equal-

ity, while a value of one reflects perfect

inequality. According to the Census

Bureau, the US Gini coefficient rose

from 0.404 in 1979 to 0.486 in 2018.

Americansdo care about this — but

not deeply. About six in 10 of those sur-

veyed in a recent Pew Research poll said

there is too much economic inequality

in the US, but only four in 10 said reduc-

ing it should be a policy priority. Soaring

consumer confidence data suggests that

most people care more about how they

are doing relative to a few years ago than

relative to the very rich.

If this is the case, Mr Trump’s re-elec-

tion campaign should benefit from the

labour market’s recent strength. Other

research suggests that many Ameri-

cans aren’t bothered as much by ine-

quality as you’d think because they

believe in the “American dream”: they,

too, can and will do better.

But Democrats are hoping to capital-

ise on concerns revealed by another Pew

poll. In 2018, only 33 per cent of Ameri-

cans expected today’s children would be

better off financially than their parents,

down from 37 per cent a year earlier. Mr

Trump’s opponents argue that despite a

low unemployment rate, the future

looks grim unless America starts to

address its inequality issues. The

November election may turn on how

well they can make that case.

The writer is a senior fellow at Harvard

Kennedy School

US election

may turn on

voters’ view of

inequality

Whatever their precise

share, the rich have been

getting richer than the rest

of America for decades

SOCIETY

Megan

Greene

MARCH 3 2020 Section:Features Time: 2/3/2020 - 19: 10 User: dana.prince Page Name: COMMENT, Part,Page,Edition: LON, 11 , 1