PsyFi Search

Saturday 29 May 2010

Laplace’s Hammer: The End of Economics

Clever Man, Stupid Idea

Simon-Pierre Laplace was a Very Clever Man who did many Very Clever Things. Unfortunately, like many clever men, having got hold of a Brilliant Idea he was rather inclined to go off and use it on everything in sight, which led to a number of Very Odd Conclusions. In fact as far as science goes, he may well have been the original man with a hammer; taking aim at every problem as though it was a nail.

As is the way of these things economists got hold of Laplace’s ideas, converted them to their own and started developing delicate and intricate webs of theories and practices. Unfortunately, over the succeeding three hundred years they’ve failed to keep up with advances in physics and biology, rather leaving economists as the only believers in an approach that suggests we have no free will, a position from which they’re having to be dug out and defused, one unexploded theorist at a time.

Newton’s Error

When Isaac Newton published his theory of gravity he knew it contained a serious flaw; his equations didn’t exactly match what was observed. He reckoned, correctly, that the differences between his theory and the observations were down to the gravitational effect on the planets of other planets – so called perturbations, which he didn’t know how to mathematically model. He said:
“But to consider simultaneously all these causes of motion and to define these motions by exact laws admitting of easy calculation exceeds, if I am not mistaken, the force of any human mind.”
And, of course, he was mistaken. The force of Laplace’s mind successfully solved the problem. Having achieved this, though, Laplace went a stage further. Because he could exactly calculate the position of any planet he could compare this with the results of observations from astronomers. What he discovered changed human history.

Observer Bias

It turned out that observers made errors, but they made them in a particular way – their observations fell about the actual position of the planet in a very distinctive pattern. This pattern, of course, was the ultra-familiar Bell curve or normal distribution. Laplace realised that human error was statistically quantifiable and, therefore, could be effectively eliminated from the data. This was a brilliant insight which encapsulated the concept of observer bias and made the observer a quantifiable part of scientific experiments.

However, having found his hammer, Laplace went looking for nails. Extrapolating from the heavens to humanity, he argued that it should be possible to predict everything, including the behaviour of people. Thus was born so-called “scientific determinism”.

Determinism in Biology and Economics

This idea infused the scientific culture of the eighteenth and nineteenth century, leading to the ideas of biological determinism espoused by Francis Galton (but rejected by Charles Darwin) and, of course, ideas of economic determinism through David Hume and John Stuart Mill. Most of the twentieth century was one long line of economists determinedly holding to the underlying concepts of determinism with perhaps only Frank Knight and John Maynard Keynes as conspicuous objectors. Drakopoulis and Torrance provide an overview of these developments and the way behavioral finance has started to address them in Causality and Determinism in Economics.

Yet even while economists were busily building physics-type models based on the assumptions of causality and the quantification of error the physicists were beginning to have a few minor, local difficulties with heat. When they tugged on the loose thread of thermodynamics, to their horror, the whole of the causality of physics unravelled before their eyes.

Entropy

The problem was that Laplace was wrong. He can be forgiven for this because he’d happened upon one of the rare areas of the physical sciences in which determinism is pretty much absolute – planetary motion. Essentially the planets sweep around the Sun in a majestically predictable fashion. Unfortunately most physics is a lot messier than this.

Perhaps this is most easily encapsulated by the Second Law of Thermodynamics which states, more or less, that everything’s descending into chaos. The order of the universe is an illusion and so determinism fails. In short order developments in quantum mechanics suggested that there are absolute limits on what we can know.

Finally Stephen Hawking showed that in the vicinity of black holes – an idea that Laplace had suggested – even the limited amount of information implied by quantum theory disappears. Which, as he also pointed out, wouldn’t matter so much if quantum mechanics didn’t suggest that everything is in the vicinity of black holes.

Gene Expression

Meanwhile biologists assumed, to the extent that they were aware of these issues at all, that quantum indeterminism didn’t really apply to them. After all, they deal in assemblages of genetic material far removed from elementary particles. However, the geneticist Motoo Kimura, one of the last century's most important evolutionary theorists then showed that DNA changes far too much to be explicable by standard Darwinian theories. In his view the driving factor behind this engine of mutation is the buzz of noise we call chaos that is the all encompassing background to the universe – gene expression too is subject to indeterminism.

Although lots of people disagreed with Kimura – and still do, often very noisily and rudely – the evidence for his ideas are growing. In 2002 a group of researchers published a paper (abstract only, I'm afraid) in which they described how they had demonstrated the effect of this background noise on gene expression. They introduced the same firefly DNA for different colours into a bunch of bacteria: if background noise plays no effect in how this DNA is expressed then all of the bacteria should look the same. Only they didn’t – the bacteria showed up with a weird and wonderful collection of different colourations, the result of random variations in the way the underlying genes were expressed: chaos in colour.

Behavioural Finance

Yet although determinism is being broken down by the slow but relentless progress of the scientific method, those pseudo-scientific subjects that have taken on the mantel of science are slow to modify their approaches. High amongst these is economics, of course. In part this is doubtless because many of the practitioners have no feeling for the history of the topic and fail to recognise the flimsy foundations of theory it’s constructed upon. But probably it’s because economists don’t understand physics.

Meanwhile the attrition of traditional economics by the closest thing it has to a scientific approach, behavioural finance, is only slowly gaining traction. Peer under the covers at any of our great financial institutions and you’re likely to find people using methods and models ultimately based on Laplacian determinism and ultimately destined for the same unfortunate end after they’ve caused a few more market crashes.

No One Knows Anything

The problem with Laplace’s hammer was that extrapolating from one situation – the regularity of planetary motion – to another – the regularity of human behaviour – is only possible through a delicate chain of assumptions and analogies. Other subjects that rely on such approaches, such as economics assuming that efficient market hypotheses which work during normal times will always apply, need to tread carefully lest they lose their way.

So here’s a paradox: although no one can know anything absolutely we can absolutely know that we can’t predict the markets and that all the people telling us that they know what’s going to happen next aren’t just know-it-alls, they’re actually know-nothings. Listen to no one who predicts anything, they’re all just victims of Laplace’s hammer.

Pierre-Simon Laplace, 1749-1827A Philosophical Essay On ProbabilitiesThe Mismeasure of Man (Revised & Expanded)

Related Articles: Econophysics, Consciousness and Cosmic Karma, Newton's Financial Crisis: The Limits of Quantification, Cyclical Growth, Form and Fibonacci

7 comments:

  1. "...we can absolutely know that we can’t predict the markets...". This is the prediction of the efficient market hypothesis. This seems to be a good prediction of economics.

    ReplyDelete
  2. although no one can know anything absolutely we can absolutely know that we can’t predict the markets and that all the people telling us that they know what’s going to happen next aren’t just know-it-alls, they’re actually know-nothings.

    I like the article. The background is super.

    It's a mystery to me how you make the jump from what you say in all the preceding paragraphs to what you say in the conclusion.

    We can't predict what's going to happen next? Why not? Lots of people have been doing it effectively for a long time. Shiller predicted this stock crash. So did Arnott. So did Asness. So did Grantham. So did Easterling.

    Do these people not count?

    You can't predict if you ignore the factors that determine what is going to happen next. I can go along with that. But there has not yet been a law passed saying that we must ignore the relevant factors (if there has been, I am in big trouble!). If we are willing to look at the relevant factors, why would we not be able to predict?

    Lots of good and smart people agree with what you are saying, Tim. I can in fairness point that out. But I can't say that I see can it. My view is that the entire purpose of stock analysis is to make effective predictions. If you cannot make effective predictions, it's all gambling, isn't it? Should we all be gambling with our retirement money. If I ever came to the conclusion that we cannot predict, I would on that day get out of stocks for good.

    In my mind, the question is: What sorts of prediction work and what sorts of predictions do not work>?

    Rob

    ReplyDelete
  3. Rob Bennett said... "Shiller predicted this stock crash. So did Arnott. So did Asness. So did Grantham. So did Easterling. Do these people not count?"
    No they don't - at least not for this reason. After all, there were lots of people making all kinds of predictions (like Bernanke's "I can't see a bubble", "it will be restricted to subprime",...). Some of them predicted a crash, therefore they seem right ex-post: so what? Bill Miller "was right" 15 years in a row, again what did that prove?

    ReplyDelete
  4. One of the biggest edges a trader can have is knowing that the future cannot be predicted, in spite of all of the obvious news laid out on the table.

    Follow price.

    ReplyDelete
  5. Unfortunately the fact that some people predicted the crash doesn't mean that it was predictable. Some people are always predicting a crash, so some people are bound to be right on the same basis that a stopped clock is right twice a day.

    So the question defaults to, why were they right? The only answer to this I've found so far I described in To Predict the Next Bust Ask An Austrian" but I don't really believe it answers the question.

    ReplyDelete
  6. So the question defaults to, why were they right?

    I agree that that's the most important question, Tim.

    Some types of predictions are sometimes right just by chance. Other types of predictions are always right because they are rooted in compelling logic. My view is that you want to avoid the former type of prediction and be sure not to avoid the latter type of prediction.

    A key point here is that not all stock predictions can be treated the same. There is a tendency among some to lump them all together and conclude either that all predictions work or that no predictions work. My view is that the key to successful stock investing is distinguishing the predictions that must work (those rooted in valuations assessments) from those that might work and might not and more often than not do not (all other types of predictions).

    One point where I think a lot of people go off track is in thinking that there is some way to invest without making any predictions. I don't see how it is possible. As you note above, even the statement "no predictions will work" is a prediction. Saying that doesn't help you avoid predictions, it just causes you to go with a type of prediction that has a worse track record than those rooted in an analytically valid analysis of the historical record.

    My sense is that predicting got a ban name in InvestoWorld because so many people engaged in the sorts of predictions that don't work that now lots of people are trying to avoid predictions altogether. But risk analysis is an exercise in prediction. If you rule out prediction, you rule out risk analysis. That can't be the right way to go (in my view!).

    Rob

    ReplyDelete
  7. Buffet is least like the other successes cited - he's *all* about survivorship bias (ie. there are say, 20k "Warren Buffet of 2030 candidates" amongst us today, but we won't *KNOW* which of them will turn out to be THE Warrren Buffet of 2030 until....2030). The others are indeed akin to broken clocks. Track record appears to be abt skill but in fact it is a lottery (which the Bill Miller example shows). However, humans NEED stuff to believe in (eg santa claus, organized religion, military basic training) which is likely a hangover from infancy/childhood where we MUST believe in our elders/caregivers if we are to survive into adolescence - we should become LESS credulous as we age but most (all?) of us fail in this endeavour. So banking/finance/econ is shot through with groupthink and logical inconsistencies but they are ones that we all agree to believe in, under the guise of rational, problem-solving, engineering, logical "best practices." Note that a century ago the best practice for banks was to be balance-sheet focused (almost completely ignoring income statements) whereas now we are probably the negative reciprocal of that. But in both cases, agreement was pretty solid all round. Carry on

    ReplyDelete