Follow the reluctant adventures in the life of a Welsh astrophysicist sent around the world for some reason, wherein I photograph potatoes and destroy galaxies in the name of science. And don't forget about my website, www.rhysy.net



Saturday, 30 November 2019

Science And The Free Market


Politics is pretty shite, isn't it ? Whether you're a left-leaning kale-munching vegan anarchcic panpsychic hipster who hates the idea of "rules", or a right-wing "lock 'em up and throw away the key if they so much listen to jazz music" type, I suspect we can all agree that politics is about as much fun as an unexpected brick to the gonads.

Science, on the other hand, is very much not shitty. Most of the hardcore, truly anti-science "it's all subjective and everything is made of wishes and moonbeams" brigade are largely confined to the realm of the internet. Admittedly the sort who think science is great except for this one thing (vaccinations/GM food/global warming/whatever) are rather more numerous, but that's another issue. Most people, having the common sense to realise that computers wouldn't work and planes wouldn't fly if our basic scientific theories weren't at least on to something, accept that science more or less gets it right in the end. It's an ongoing struggle to explain the finer details, but this fundamental point seems to be widely accepted.

So why is this ? Why does science seem to be high performance whereas so many other institutions seem to be at best dysfunctional, at worst outright harmful ? How does science manage to establish objective truth so well while politicians seem to prefer to actively ignore it ? Why do scientists actively enjoy having their theories overturned while a politician would sooner walk off a cliff than admit to any imperfections ? Surely, if we could only apply these lessons from science to other aspects of society, we would do nothing except make the world a happier, nicer place.

A confession : this post is a sort of preamble. When I thought I was about to finish, I suddenly realised that I had all the tools to forge a better political system starting me in the face, and I could hardly let that go. So in this post I'll look mainly at the contributing but less important factors : the people involved and the networks they live in, why science has some intriguing similarities to the free market so beloved by conservatives, and I'll emphasise that not everything in the world is quite so bad as it seems (though it is pretty bad).

But mainly this post only exists because it was necessary to write it in order to figure out part two, which is much more interesting. There I'll look at what really makes science work and present a proposal for a parliamentary system designed to Make Politics Work Again. My limitation here is that I'm going to focus on the scientific/political processes within their own systems, not within the wider context of the public/media perception. These are important, probably more so than the systems themselves, but in fairness restructuring the whole of society is a wee bit much to bite off in one go.


What is it that science does well ?

There's a nice, very tongue-in-cheek chart in Niall Ferguson's broken masterpiece on societal networks :


I particularly like the one for Microsoft, where lots of different departments are pointing guns at each other. It's very silly, but it makes a serious point. In any organisation, not only is it important which individuals interact, but the nature of that interaction is also critical. People can see each other as peers, rivals, underlings, authorities, casual acquaintances, sex objects, doormats, etc. They can also assess each other as competent, stupid, unprofessional, hilariously smelly, moronic, reliable, and a host of other weird and wonderful personality traits.

Both the structure and behaviour of scientific institutions matters to their success. Key to this is that mistakes are absolutely integral to the process - it is literally true that it's not research if you know what you are doing; failure is always an option. I also spit in disgust on the notion that progress is driven by lone revolutionary geniuses; no man is an island, and while there are certainly geniuses, virtually all of them stand on the shoulders of people of about average height, who often go around doing really dumb things.

Yet because, rather than in spite of, this apparently perverse predilection to accept mistakes with good grace, science is astonishingly successful - within certain limits. When it comes to analysing the data at hand, the consensus viewpoint seems to be feckin' awesome at accepting the best, most rational viewpoint the current data and methodology suggests. That sentence has been delicately phrased : to change one's mind based on new evidence is no bad thing, and once upon a time it was entirely sensible to consider the Earth to be flat and genuinely brainless to suggest that it was round.


I call this the efficient consensus hypothesis. That is : the current consensus gets pretty close to (though rarely actually reaches) the most rational conclusion currently possible. For today's exercise, that's how I shall define truth, rather than using a more absolute standard; strictly speaking science does not describe truth, if there even is such a thing. I shall ignore deeper philosophical questions as to what we mean by rationality and whether it's always a good thing. It's anyway clear to anyone with an ounce of sanity that politics is not nearly rational enough, and that's my main target for reform today.

Let me break that claim down, just to be even more emphatic. I apologise to anyone who thinks I'm labouring the point, but I've learned that the internet is full of unnecessary pedantry and the wilfully stupid, so it pays to be clear.
  • Scientists are good at doing science. They are not necessarily also amazing basketball players or fashion designers or zookeepers, although some of them are. They're good at analysing the data and solving the problems put before them : I will here make no claims regarding their broader critical thinking skills whatsoever. I will not even comment on their abilities as predictors, only their capacity for analysing current data - not for accounting for how conditions and circumstances may change in the future and how different factors may interact in novel ways. 
  • The scientific consensus is strongly, though not perfectly, robust against conclusions which defy the current evidence and analytical methods. It does not preclude individual researchers from being lunatics (indeed, some have it that a consensus without any dissent is a sign of systemic, groupthink bias). It's the capacity of the group to establish truth I want to look at here, not the individual elements.
  • The consensus view generally changes rapidly and correctly in accordance with the available evidence. It is imperfect and incomplete, but ultimately successful. For example, there was a time not so long ago when science was able to show how an aeroplane flies while simultaneously having nothing much to say about how to design a nuclear bomb.
The key thing to remember is that I'm simply claiming that modern science (say, as done in the last couple of centuries or so) rapidly accepts the best conclusions currently possible, not that it never makes mistakes - in fact it can't advance at all without making mistakes. In my view, whenever one looks at just about any claim that scientists could have done better, one finds that this is nonsense given proper context. Science continuously improves not just its findings but its methods as well - finding mistakes in those methods is part of the process. I think it is rare indeed that the scientific consensus has ever been at odds with reality when there was good evidence / better methodology staring it in the face, at least for any significant length of time.

How, then, does science accomplish this amazing feat ?


It's because scientists are just better people than everyone else, obviously

"I'm, too snooty for my shirt, so snooty it hurts..."
This is the hope of those wanting to see more scientists in politics. I previously went on an extended rant about this, concluding that this wouldn't work without wholesale change to the system (next time I'm at last going to at last suggest what sort of wholesale change that might be).

But I suppose in principle it could be that scientists are all innately more curious and interested in the truth than other people - maybe the network stuff is just an emergent consequence of scientists' natures rather than a cause of it. But even then, there would have to be something about academia that does an outstanding job of selecting such people in the first place, so there would still be valuable lessons to learn. And the network structure would still matter, since in order for opinions to change rapidly, there must be efficient communication between all interested parties.

But I don't find selection a terribly convincing explanation. Scientists have diverse personalities and interests, even if there is (maybe) a greater tendency to prefer staying in binge watching science fiction shows and avoid talking to anyone more lively than a cactus. I've met people who are genuinely interested in the truth and others who, methinks, doth protest too much, who clothe their agenda by loudly insisting on how wonderfully unbiased they are. There are those who are exceptionally, almost unbelievably well-informed about the entire state of their field and others with more limited knowledge. And outside of their field of expertise, plenty of scientists have some damn fool ideas about the world.

Moreover, the desire to keep learning can be found in people of all walks of life. It's not only scientists who like documentaries or read the news or change their minds. True, some people only enjoy learning things which they're already prepared to accept, but still many non-scientists can delight in learning things which contradict what they previously believed. At least in some circumstances, at any rate.

On the other hand, I should also mention that I simply don't buy the popular suggestion that everyone is a natural-born scientist and the education system drums it out of them : I think all youngsters are curious simply because they don't know very much and know they don't know very much. They've no choice but to be curious, otherwise they wouldn't learn anything and natural selection would quickly take over. So it seems to me that selection plays only a limited role. Scientists don't have a monopoly on reasonably-minded people, nor are they exclusively sensible themselves : sensible people enter, endure, and are perhaps created by other networks besides science. Something else is going on.


Maybe it's more about the system scientists operate in than the people themselves

After all, it's easy to be a saint in paradise. Unfortunately no-one has yet extended this to the logical extreme of giving everyone MOAAARR PUPPIES.
I've mentioned this before a few times, most explicitly when comparing science to politics, and why we shouldn't expect scientists to fare very well as politicians. There are many different reasons for this. For one, scientists are trained to analyse the facts when forming a conclusion, without regard for people's feelings - if they did account for how people would perceive the facts, they'd be lousy scientists. Well, except for the social sciences, obviously.

But scientists also exist in an environment which supports this process. In contrast, partisan political systems are explicitly designed to give people a choice regardless of the facts, so the opposition parties criticise the government almost regardless of what they do (and the media constantly attack literally everyone as though they were worse than Hitler). There's no reason at all to think that scientists would do especially well in that kind of atmosphere. This is also, perhaps, why ex-military figures traditionally don't do well in politics either. I've also described how the routine, ludicrously extreme criticism by the media renders political bullshitting all but inevitable - no matter how honest and well-intentioned any politician may be to begin with.

Another perspective is to consider how science arrives at a consensus. That is, it doesn't hold a vote and decide, "right lads, we'll assume this is true and move on.". Rather the consensus is the emergent majority viewpoint : it's what most people happen to believe after quasi-independently examining different ideas, but at no point is a vote required or taken.

This sort of quasi-independent examination appears to be rare indeed in politics. Politicians are not even allowed to express their own thoughts, much less pursue investigations at their own direction. If a scientist says something unpopular with other scientists, they'll have to fight to defend their idea but that's about the extent of it; if a politician says something voters and/or their party machinery disapproves of, they're liable to lose their job. There's a much, much greater constraint for a politician to say things people already want to hear and do what the rest of their party tells them. The scientific network is a lot looser, far less threatening, and deliberately more tolerant of dissent than political entities.


A balancing act

The balancing act often fails when it comes to individual scientists, but, as we'll see a bit later on, not for the system as a whole. 
Tolerance of dissent and a loose, fairly egalitarian network make it easy to explain why science doesn't fall victim to a false consensus - that is, a majority viewpoint held simply because it's already the majority viewpoint, not because the evidence is actually compelling. As well as individual freedoms (not afforded to politicians) I've suggested that this is avoided because of the competitive collaborations that make up scientific institutes around the world. Each has a vested interest in disproving the other, with the consensus emerging from both collaboration and competition between different groups.

Individual researchers experience two competing motivations. On one hand, they try and publish as many papers as possible because this is good for their career - which naturally tends to encourage salami publishing of very incremental, mediocre results of little importance and controversy. That tends towards at best slowing progress, and worst stalling it. But on the other hand, researchers need to publish papers which are as radical as possible, because if they can convince people they've made an astonishing new discovery then that's a) innately interesting and b) very good for one's career indeed.

Fortunately the whole thing is tempered by the oversight of peer review, which prevents authors from making wild, unjustified claims about the healing power of pugs or whatever. Peer review acts against the tendency to publish radical, breakthrough results, whereas the need to amass high numbers of publications and citations pushes back against this - and peer review pushes back again by saying, "nope, that result is just too dull". By and large, papers are published which are not such tiny incremental advances as to tell you nothing at all, nor so radical as to be purely speculative and unsubstantiated. Like everything else, it's a highly imperfect system, but it functions sufficiently well to accomplish its main goal. More on peer review later, and especially in part two.


Science : a Libertarian dream ?

"Let everyone do what seems best to them" is not something that one normally associates with science, but the comparison is not completely crazy.
But hang on, all this is quite remarkable. "Competitive collaborations" describe just about anything. Farmers collaborate on their own farms but compete with each other. Tech companies collaborate internally but compete with each other. Bankers, fishermen, politicians*... it's pretty much the same across the board. And yet science avoids the perils of monopolies that plague the free market without having to resort to external regulation; it also largely avoids the destructive forces of tribalism that can obliterate sensible decision-making in politics. How in the world does it do this ?

* With strict caveats - more in part two.

Well, first, let's not go nuts with the criticism of non-science organisations. Many other institutions do also accept and discover at least some fundamental truths; some aspects of virtually all systems are successful otherwise they wouldn't last very long at all. The point here to highlight the differences between what works and what doesn't : scientists don't act scientifically when they go for a poop or cuddle a kitten; bankers don't act with ruthless self-interest all the time. Probably.

Certainly individual scientists and institutions can be persistently belligerent and foolish, yet the scientific system as a whole is robust to effects that seem - at least on occasion, definitely not always - to cripple other institutions. Or perhaps it is not so much immune as it is disease tolerant, when infections still spread but their effects are rendered impotent. That would be quite an elegant explanation as to why scientists often seem like perfectly normal people - with all the imperfections that implies - but the edifice of science as a whole produces results far greater than the sum of its often deeply flawed parts.

Still, science does have an uncanny resemblance to a Libertarian Utopia. Very few scientists indeed ever interact with government officials, and in any sensible country it's scientists who set their own agenda, not the government. As far as research goes scientists regulate themselves : peer review is a fantastic example of self-regulation that actually works. Of course the tremendous irony is that most research is government-funded, and wouldn't happen at all but for taxation. So perhaps the true Libertarian ideal is one where the government shuts up and gives people money.


This is not to say that scientists operate in complete ivory-tower independence. Different institutes operate in markedly different ways. And of course there is an interface between scientists and government, with research councils acting as buffer to prevent direct government interference. But there's little enough direct government meddling in most day-to-day research operations.

If I had to sketch out the global scientific network, I'd probably say it's something like this :

A rough concept sketch. Some institutes have very different external and internal connections, though most universities tend to be similar. Of course, the connections depend on specific research field for each institution.
Individual institutions, in my (both direct and second-hand) experience, can function quite differently to one another. Major research facilities tend to be extremely hierarchical, with strong top-down control. This is fine if your goal is to examine one single thing in one single way, e.g. CERN's search for the Higgs Boson. Smaller facilities, and most universities, tend to be more chaotic "small world" networks, where lots of people know each other, but not so much as to enforce a single viewpoint. This is much better for creativity but worse for converging on a single solution (which is precisely what gives the consensus such value - it's very hard won !).

On larger scales, connections between different institutes tend to be sparse but important; usually there are a few key individuals (older, established professors) with enormous networks, though only the most junior researchers will have no connections to other places at all. I suggest that this too is dense enough to facilitate rapid communication, but not so tightly-knit as to enforce a global consensus.

Mind you, this can only be an intuition-based sketch. Whether there are significant differences between how science and other societal institutions are organised is something I cannot properly address, though I'd very much like to see someone with network analysis experience tackle this properly. At face value it would seem that this "competitive collaboration" model isn't enough to explain why science works while other areas are dysfunctional. So I shall say no more about the network structure : it probably does matter, but it may not be the key to success. The network, I think, is in this case more a consequence than a cause of the methods scientists have adopted.


The end is not yet



We should also recognise that while other systems have problems, we should not think they are doomed. The liberal, social democracies of the West have endured enormous turbulence. While I think it's credible to suggest they're under threat, it would be a mistake to pronounce the death of any unstable system. And it's taken tremendous pressure even to bring them to their current crisis. Similarly, rumours of the death of capitalism have been perpetuated since at least the 19th century, but it's still not dead. It too has endured plenty of crises yet no credible alternatives have ever really been devised.

But it's not so much that any of these systems might be on the verge of collapse that's interesting so much that it is that they have problems which science simply doesn't. Problems of market forces creating monopolies, price fixing, and, more chronically, enormous wealth inequality and skyrocketing prices, are all real problems but solvable through regulation. No sensible person would say that the free market means it should be anything other than free in the sense of freedom under law. Democracies have problems of voters making astonishingly stupid choices; the free market of the media has led to comically absurd levels of polarisation and often outright hatred.

The point is that while economics and politics both seem to need continuous management, a guiding and hardly invisible hand of the state, science doesn't. Why ? And in particular, is the popular "efficient democracy hypothesis" even credible ?

I'll suggest on this latter point that no, it isn't. It's true that "everything's getting worse" is a popular fallacy, and by many statistical measures things are getting progressively better. Fair enough, but it would be stupid to then presume that politics must be doing anything remotely like close to optimal given the available options, let alone that the current crises are mere aberrations. I'll be blunt : anyone still supporting Brexit and/or Trump is an idiot, and most of those who supported them to begin with (especially Trump) were idiots anyway. The major reason voters didn't make such stunningly awful choices in the past is simply because they weren't given the opportunity to do so; let's not forget, of course, that politics has made mistakes the like of which modern science could scarcely believe.

And I'll also add that politics might in some way be a victim of its own success. In the past, when there were clear and present external dangers, nations have often rallied to defend themselves (though by no means always with success). But when a country is prosperous and perceives little external dangers, it falls into complacency. Bereft of outside enemies, the system has a horrible tendency to create new ones of its own people.

Okay, rant over. What has all this preamble taught us, if anything ?


Summary

Science does seem to do pretty well at selecting people who are able to do science. Since that does require some degree of both critical and analytical thinking, it would be a mistake to say that scientists are a good representative sample of the population. But it would be foolish to say that this is the only or even main factor at work : plenty of other people are intelligent, rational, and all-round good eggs. Lots of people enjoy learning and even being wrong, but other sectors seem to have problems that science doesn't.

The network also seems to be an important contributing factor but it can't be the whole story. Science balances competition and collaboration very well, but other areas try and do this but often - but by no means always - either fail or cause everyone massive problems. That's not to say that science and technology don't also cause problems too (the Industrial Revolution transformed the economy but also wrecked the planet), but those, for the most part, would seem to be a result of improper use rather than innate problems of research itself.

Make no mistake : individual scientists can be thick as shit. Yet the overall consensus always seems to do pretty well given the evidence available, whereas politics struggles to see the blindingly obvious. As for unchecked market forces, it's not that they never do anything good so much as they don't seem to give a damn about some pretty horrific side effects.

If it's not the people, and it's not the network, where else could the secret be hiding ? My guess is it's the methodology scientists have adopted. Next time I'll look at that in detail and try and see if we can apply this to other areas.

No comments:

Post a Comment