Human fallibility and robust institutions

Human fallibility and robust institutions

Note: I began writing this post before the election; I finished it after. I’ve indicated where the break is.

It’s amazing that a system of a billion weasels trying to hose each other, i.e. capitalism, works as well as it does.

Scott Adams, creator of Dilbert.

Many human institutions are not optimised for perfection. They’re optimised for robustness. Capitalism is a case in point. Certainly, a huge amount of human endeavour is wasted in an arms-race of advertising, or in Shumpeter’s ‘creative destruction’. But the alternative, a planned economy in which state experts determine how much should be made, where it should go and what price it should be consistently fails. Firstly, it’s all but impossible for those in command to know what to do as well as the invisible hand of the market; secondly, and perhaps more importantly, a command economy creates a single point of failure for corruption and deliberate control. The billion weasels triumphed and the West defeated the Soviet Union in the Cold War(1).

Similar choices can be seen in the most effective and most just systems for.law and government. Just as one could argue that – given a perfect human in charge – a command economy would avoid the inefficiencies of the market, so one can argue that it would be more efficient to have a single judge investigating a case, rather than prosecutor and defence lawyers trying their hardest to outwit each other, make each other look bad and dig out inconvenient facts. Or couldn’t we do away with all the lies and bad mouthing and tribalism in politics and just do what the experts think is right?

But these solutions suffer from the same failure. Each side in a court case having their own spokesperson isn’t perfect, even when the resources on each side are equal. But having a single person in charge of administering justice is worse. Even if they’re honest and unbiased, there wouldn’t be the incentive there is currently on both sides to really dig out the evidence precedents and so forth to understand the case fully – see the classic study of how judges are more likely to grant parole depending on whether or not they’ve eaten lunch [Edit: it appears this study has failed to replicate. But the point that judges are fallible remains true!]. But more seriously, there’d be no checks on discrimination, bias or abuse of justice.

Similarly, in politics, we recognise that many of those who want power will be ambitious and liars, and rely on a system in which their opponents and the free press are able to call out their errors. As Churchill said, it’s the worst form of government, apart from all the others.

It’s no surprise that lawyers and politicians are traditionally two of the most despised professions. But the systems work not in spite of human fallibility, but because of them, channelling self interest, ambition and, yes, weaseldom into positive societal outcomes. As Adam Smith said:

It is not from the benevolence of the butcher, the brewer, or the baker that we expect our dinner, but from their regard to their own self-interest.

Adam Smith, The Wealth of Nations

Of course, we do make some interventions. We break up monopolies; we provide legal aid so people who couldn’t afford one have a lawyer; we put limits on electoral spending. But these are typically highly circumscribed, objective decisions. There is less space for bias and abuse of power if one is determining a monopoly than saying which product is best; to providing lawyer than determining who is innocent; to placing a limit on spending than to setting out which views can and cannot be expressed. We should not create single points of failure.

Richard Feynman argued that science was effective not because scientists were intrinsically better than other people or because science never made mistakes. Rather it was because the process of science had a mechanism to correct mistakes, which made it robust to human fallibility.

We’ve learned from experience that the truth will out.  Other experimenters will repeat your experiment and find out whether you were wrong or right.  Nature’s phenomena will agree or they’ll disagree with your theory.  And, although you may gain some temporary fame and excitement, you will not gain a good reputation as a scientist if you haven’t tried to be very careful in this kind of work.

Cargo Cult Science, Richard Feynman (1974)

Robust processes – in science, in economics, in law, in government – aren’t optimised to prevent mistakes, because that’s impossible. They aren’t optimised on the basis that human beings are morally perfect or incorruptible, for they’re not. Rather they are robust through being self-correcting, by being able to recover from errors and by being able to operate in a system in which individual actors in it are imperfect.

In modern times, many people are tempted to short-circuit these processes, trading robustness for efficiency. It’s understandable, when we look at the inefficiencies of the processes, as well as the character of some of those engaged in them, on all sides. But it’s not the right answer.

Why don’t we, you hear people say, put this process outside of the political realm, so the experts can make the decisions they think best? Maybe we should set up a body to rule on what politicians are allowed to say, or which parties are allowed to run. Maybe an independent commission to vet candidates, to ensure that only those who have values compatible with the European Convention on Human Rights – or are suitable loyal to Britain and British Values – should be allowed to run> Or how about giving massive, unaccountable, multinational tech companies that avoid paying tax the unilateral power to vet what views are allowed to be expressed?(2) That would soon set things to rights. Well maybe it would, for a short time. But as Galadriel said to Sam Gamgee, it would not end there.

[The below was written after the election happened. I don’t think it fundamentally changed the message of what I was going to write, but the ending clearly reflects a greater degree of outcome knowledge].

For power corrupts, and absolute power corrupts absolutely. Give a body the power to vet politicians and sooner or later it will be working for the status quo and the interests of the rich and powerful. Do you doubt that? Just think how easily it would be to disqualify Corbyn under ‘objective’ grounds. Or look at the history of the Suffragettes, the Protestants, the Civil Rights Movement, the Labour Movement or any other movement you care about to see how those in history who challenged consensus wisdom were called ‘extreme’, ‘dangerous’ or had rights taken away from them.

Give a system a single point of failure and it will fail. Never assume a single person, or group of people, has exclusive access to the truth and exclusive power to enforce it. In politics, this leads us to Hitler, Stalin and Mao (or, less dramatically, but still very much to avoid: Pinochet, Marcos and Castro). In justice it gives us the Spanish Inquisition, Star Chambers and the Cultural Revolution. Trying to perfect human nature, to match it to the perfect system is, historically, equally bad and leads to atrocities such as eugenics and reeducation camps.

Everyone has a right to speak. Everyone has a right to justice. Everyone has a right to vote. And those rights must be from the full panoply available, not circumscribed or chosen from a predetermined list approved by those in power.

The mistake some people make is thinking that democracy is about the little things. We don’t go to the people to find out the precise cost of a hospital: that’s what technocrats are for. What democracy provides is the overall direction: the free market or socialism; higher taxes or lower ones; immigration up or down; in the EU or out of it? And most important of all – and what people are best at knowing – have we had enough of those currently in office, and do we need to vote them out? The fact that every government falls eventually may be the ultimate validation of democracy.

No system is perfect. The fall-out post Brexit highlighted a number of weaknesses in our constitution, most notably the fact that it was hard to deal with a situation where the majority of people outside the House of Commons had voted for something that the majority of people inside did not want to implement (a situation that no longer applies). But when we consider strengthening it we must do so in a way that makes it more robust, not less; that avoids creating vulnerabilities and single points of failure; and that recognises that humanity is both noble and fallible: standing, as Pratchett put it, “Between the falling angel and the rising ape

Ultimately, there will always be disagreement over whether or not we should have had to vote again to deliver Brexit – and different views over the outcome. But, as a nation, we should take pride in the robustness of our democratic system that meant the gridlock and division was settled not in the courts, not in the streets, but by another democratic vote.

(1) There are other reasons to prefer free markets, e.g. the individual freedom to start a business, sell one’s labour or buy the goods and services one wants, but it was the fact that it worked better, as an economic system, that was decisive in winning the Cold War.

(2) No, I don’t understand this one either.

2 thoughts on “Human fallibility and robust institutions

  1. Good article. I find this kind of examination of emergent systems, and how surprisingly well they work, very interesting.

    Minor quibble, though: the judges/lunch study hasn’t stood up to scrutiny, and another proposed explanation is that defendants without representation are scheduled at the end of sessions (i.e. just before lunch) and also tend to receive harsher sentences (because they’re not represented, not because the judges are hungry).
    http://nautil.us/blog/impossibly-hungry-judges
    annieduke.com/no-judges-dont-give-harsher-sentences-hungry-annies-newsletter-october-5-2018/

Comments are closed.

Comments are closed.
%d bloggers like this: