black-swan1

If we take Taleb seriously

To take Nassim Nicholas Taleb seriously one first needs to unsubscribe from his Twitter feed, then un-like him on Facebook and then read his books, especially Antifragile and Black Swan: his daily output is somewhat erratic, whereas his books are quite profound.

Of course I can’t summarise all his views, but the idea I want to look at here is based on the following chain of points from his various writings.

  1. We don’t know as much as we think we do (cognitive biases).
  2. We can’t ever know as much as we’d want (only have prior information which is not necessarily a guide to the future).
  3. For these reasons we’re very bad at planning for low-probability but high-impact events  — these are black swans.
  4. Therefore we should aim to minimise our exposure to black swans not by fooling ourselves into thinking we can calculate their likelihood or magnitude, but by insulating ourselves against large systemic risks (the goal of finance reform).
  5. This may mean taking a bunch of small hits to avoid getting a big one (evolution, immunity, building muscles follow this logic).
  6. It may mean simply recognising that some systems are too complicated for us to understand so we shouldn’t intervene unless it’s critical (Iraq, the human body, ecosystems).
  7. Ideally we make ourselves more than merely resilient to shocks, we can in fact benefit from shocks by becoming antifragile.

That’s probably enough for now.

Suffice it to say that I think Taleb’s general scepticism (he calls himself a “skeptical empiricist”) along with his call for recognising our susceptibility to black swans is an incredibly sane way to look at the world, from a somewhat insane mane. Cue Dryden:

Great wits are sure to madness near alli’d;

And thin partitions do their bounds divide

Anyway, many agree with his analysis, viz a vis the banking system, where we are continually exposing ourselves to massive downside risks, with too-big-to-fail corporations unable to account for highly unlikely events and succumbing to the false security of increasingly complex models based on the fatuous attractiveness of large datasets (one of his best points is about what he calls the Ludic fallacy). Many would therefore agree that we should default to accepting small risks to avoid a very large risk, buy limiting the size of corporations, say, or creating legislation that forbids future bailing out of banks, etc.

I think we should take this approach further and look at personal moral choices and the larger problems the world is facing, all through the lens of Taleb’s sceptical empiricism and large-risk aversion.

First off, climate change is an obvious case in which we are collectively assigning a low probability but high risk profile to the eventuation of catastrophic climate disruption. Even if we could somehow prove that it’s relatively unlikely that the results of producing a lot of greenhouse gases will be catastrophic, it’s still saner to minimise meddling in a large, incredibly complex system (the biosphere) that we don’t understand and which we don’t know how to fix.

But what about those moral issues on which you’re somewhat undecided? (Indeed I hope that’s most moral issues, because if you’re perfectly confident in all your moral convictions you’re probably a fundamentalist of some kind.) Eating animals, for example, is an issue on which I was highly ambivalent, but always erred on the side of eating meat and thereby maintaining the delicious status quo. When I thought more about Taleb’s ideas last year, though, I decided that the downside of being wrong (according to some future, better informed self) in one case was much worse than the other; namely, if it really is OK to eat meat (according to a complicated set of ethical arguments and taking into account the usual points on each side — we all know them, I obviously can’t go into it here) then by becoming vegetarian I will have missed out on thousands of tasty meat meals; but if it isn’t OK and I keep eating meat I will have contributed significantly to the amount of cruelty and suffering that my future self would wish I had repudiated. The thought of possibly doing something quite seriously morally reprehensible seems like a much greater risk than missing out on some food. So now I’m a vegetarian.

Abortion is something most progressives are in favour of and even many conservatives favour abortion in particular circumstances (like following rape or in the event of severe congenital problems with the foetus) but generally even pro-choicers aren’t in favour of aborting babies once they’re born — that is, of course, there is a time limit to the abortion, a threshold in foetal development beyond which the unborn baby’s rights trump the mother’s. Some jurisdictions allow up to 25 weeks, others 22 weeks, etc. Assuming there is some time during development when it is OK to terminate a pregnancy and some time when it isn’t, perhaps we shouldn’t be so keen to push it to the limit and instead favour a margin-of-error approach so that we’re not terminating pregnancies we shouldn’t be.

This is a strange way to reason, I know, but I think it might have some sense — even though the foregoing paragraph probably alienated me from most of the Left ilicet (right away); and those Right or Left who remained, probably left with the Latin.

Another example is GM crops. I’d always seen opposition to this as a kind of cutting off your improved nutrition and crop yield nose, to spite your anti-corporate or anti-scientific face: the benefits to poor and even rich people seemed to outweigh, significantly, the profiteering and sleaziness of Monsanto et al. But perhaps mucking around with the genomes of plants in our food chain is a terrible idea? The case would be made not on superstitious, don’t meddle with mother nature grounds, well, not exactly; it’s more that the science suggests properly tested GM crops are fine, but that there may be environmental unknown unknowns which naturally selected crops would have surmounted. This sounds anti-scientific, now, except that science teaches us nothing if not the danger of hidden variables, the limitations of induction and the inherent uncertainty in all knowledge. Maybe.

Finally, the paradigm case is nuclear weapons. The systems are still in place, even in the aftermath of the Cold War, and although the risk of false or real alerts may have reduced, it is still present and the ludic fallacy of game theoretic endorsements of nuclear deterrence or the MAD doctrine combined with the cognitive bias against highly unlikely catastrophic events (not to mention the cognitive dissonance of those living immediately downstream of the dam that could break, cf. Jared Diamond’s Collapse) mean that the nuclear threat is a mammoth black swan we already know about.

So what? Go veg, no late term abortions, ban nukes, reduce emissions and limit GM crops… quite a mishmash of positions that don’t really align with any established ideological camp. This gives me some encouragement and leads me to suspect that this might be a novel way to try and approach moral and ethical questions without succumbing to the usual habit of forming an emotional opinion and then merely finding all sorts of cunning post facto rationalisations. We’ll always be able to rationalise a position so instead I’m trying to prescind from the frail human biases (like my all too human proclivity for bacon) and the arbitrariness of ideological clustering (death penalty unacceptable, abortion OK). Early anecdotal evidence suggests that people hate this method of reasoning, but two points are worth making about that: anecdotal evidence is an empirical vacuum and one of the most pervasive cognitive biases is the one which makes people extremely disinclined to recognise and accept their own cognitive biases.