Narrow conceptions of determinism

Still haven't had a chance to read Taleb's Black Swan as my local bookshop hasn't received it yet but I did read, last week, his earlier (and entertaining) Fooled By Randomness which is pretty eye-opening by itself. In passing, I was reminded of Popper's stance on determinism - one of the few things I reckon Popper got wrong - and it occurred to me that much of the discussion around determinism is hampered by the fact that the popular conception of determinism, including that of Taleb and Popper, is overly narrow. Taleb makes the correct point that we often mis-identify predictable patterns in mere "randomness". But one needn't posit "true" randomness under indeterminism for this to be true. Even under determinism, any complex system is going to be "functionally" random with causes "effectively" (but not "in principle") impossible to identify. It seems to me that Taleb and Popper try (and fail) to establish an "in principle" objection to determinism but an "in principle" objection to determinism is not necessary to show that "naive determinism" - the idea that simple cause and effect are easy to identify and can be used to make accurate predictions - is wrong. In other words, the problem with Laplace's Demon is not that it would be impossible for such a "vast intellect" to predict the future but that naive determinists vastly underestimate (and misunderstand) just how (unimaginably!) vast that intellect would have to be to process the amount of information required.

Share this

"Randomness" isn't even a

"Randomness" isn't even a terribly definable concept in the absence of limited knowledge. A retrospective view of a random phenomenon can always find patterns; what defines randomness is that the information about those patterns is not knowable in advance.

The alternative to determinism, then, may be in a sense randomness, but in a larger and more fundamental way is simply incomplete information, fallibilism, limits on certainty.

To try to tease out randomness from predictability in the absence of the knowledge necessary to make those predictions is to be confused about the nature of randomness.

I do understand that random

I do understand that random phenomena exhibit apparent patterns and one of the points Taleb makes is that we are wired to read too much into these and to assume we can make predictions based on such patterns. The point I'm trying to make is that it is an error to conclude that what we observe as "randomness" is proof of indeterminism versus determinism (as opposed to unpredictability versus predictability). I'm saying that the universe probably is determinist but remains just as unpredictable. The barrier to deriving patterns and making predictions arise from complexity (even under very simple systems) rather than "true" (indeterminate) randomness.

Certain definitions of determinism are unfalsifiable.

I'm of the opinion that if randomness is part of the definition of determinism then the concept is unfalsifiable. This is for exactly the reasons you give. I don't know how you can refute the claim "It's not random just so complex that you aren't smart enough to figure it out".

Concepts of randomness

There are more than one concept of randomness. One concept of randomness concerns how a series was generated. For example, a truly random generator could in theory generate 100 heads-up in a row (in, say, its first and only run after the generator was created and before it was destroyed). It would still be random, because of how it was generated.

But another concept of randomness is Kolmogorov randomness (intuitively, the basic idea is incompressibility). It does not concern how the series was generated, but only how the same series could be deterministically generated. For example, if you flip a coin 100 times and, purely by luck, it just happens to land heads-up every time, then the sequence is not random in the Kolmogorov sense, because the same sequence could be deterministically generated by a minimally complex program. The string has very low, I imagine minimum possible Kolmogorov complexity

Don't quote me on this, but my guess is that the vast majority of strings of any given length are Kolmogorov random (i.e. the deterministic generator would be larger than the string itself), so that the probability is very high (though not exactly 1) that any given string if it was generated randomly is also Kolmogorov random.

I think, unfortunately, that the Kolmogorov concepts of complexity and randomness are of more theoretical interest than practical interest, because of the difficulty (maybe impossibility) of computing the K. complexity of any given series. However, it does allow me to define a deterministic true (not pseudo) K. random series generator as one which produces one long K. random series through the lifetime of its use. The deterministic true K. random series generator of course needs to be at least as "big" as the aggregate series that it is expected to generate in its lifetime (because of the definition of K. randomness). The payoff of my discussion is this: the world that we live in, even if it is deterministic, may in effect be such a K. random series generator in certain respects (e.g. in generating market data) because of its massive size and internal complexity.

By the way I bought Fooled by Randomness a year ago without reading it but pulled it out of my library last week or so because you mentioned it (I think that was you), and I liked it enough that I ordered and received The Black Swan today from Amazon. So thanks for the tip. He's got a bit of an attitude that the reader may want to filter through, but otherwise it's a nice review with interesting applications of clashes between human intuition and reality.

He certainly does have a bit

He certainly does have a bit of attitude! and might well be insufferable in actual R/L company but you're quite right: clashes between human intuition and reality.

The payoff of my discussion is this: the world that we live in, even if it is deterministic, may in effect be such a K. random series generator in certain respects (e.g. in generating market data) because of its massive size and internal complexity.

Yes, this is really the type of thing I'm getting at. For all intents and purposes the world we live in is "random" but the error is in concluding from this that the world we live in is indeterminist. I'm sure there's a specific logical fallacy here* and, ironically, I'm pretty sure that Taleb refers to it in his book!

* Ah that's it: Affirming The Consequent

Actually I think you may

Actually I think you may just be misreading Taleb (though not Popper), which would be his fault more than yours since it's pretty easy to do. It's been a while since I read FBR, but I remember Taleb stating (in more than one place IIRC) that his definition of randomness is basically isomorphic with incomplete information. He works at the epistemological level and I don't *think* he's making any metaphysical claims, though one could easily get that impression because he's often sloppy with his words.

I'm only basing this on

I'm only basing this on Fooled by Randomness as I haven't read Black Swans yet so maybe he clarifies his view in the latter but my strong impression was that at least some of his endorsement of Popper was based on Popper's view on determinism and what I call "naive determinism" does seem to be a popularly held caricature of determinism. Perhaps I was reading too much into it.