Nassim Taleb on the black swans of war

[MadImmortalMan]

No Joan.  He said *their* statistical tools don't work.

Referring to all "social scientists" !
It's not like physicists have some magic set of statistical tools that no one else knows about or uses.

No, it's that the social scientists have a different goal, and excluding the outliers makes it more likely that their work will be useful. I don't think that's necessarily a criticism of their work. They're trying to find out things like who will win an election, not prepare for random events.

[garbon]

Someone had an incredibly large sig.

[Ed Anger]

I wish.

[The Minsky Moment]

No, it's that the social scientists have a different goal, and excluding the outliers makes it more likely that their work will be useful. I don't think that's necessarily a criticism of their work. They're trying to find out things like who will win an election, not prepare for random events.

Social scientists is such a broad category that it makes so sense to say that.  Some social scientists focus on outliers.  Like for example, Talib.  Who happens to be a social scientist.

[frunk]

There are statistical tools for dealing with non-normal distributions, including fat-tailed distributions.  So Talib is just wrong.  He didn't say normal distributions were more tractable, or easier to deal with, or cet. par. require less data to reach significant conclusions, all which would be justifiable.  He said statistical tools "do not work," no qualification.

The key bit is "Their statistical tools".  He's assuming (correctly in many situations) that they aren't using the tools that can deal with some non-normal distributions.

[DGuller]

There are statistical tools for dealing with non-normal distributions, including fat-tailed distributions.  So Talib is just wrong.  He didn't say normal distributions were more tractable, or easier to deal with, or cet. par. require less data to reach significant conclusions, all which would be justifiable.  He said statistical tools "do not work," no qualification.

The key bit is "Their statistical tools".  He's assuming (correctly in many situations) that they aren't using the tools that can deal with some non-normal distributions.

That was my interpretation as well, and I think Joan is reaching a little in his interpretation.  If Taleb believes that there are no statistical tools to deal with fat tails, then my guess would be that he wouldn't publish a paper where he uses statistical tools that deal with fat tails.

[Warspite]

There are statistical tools for dealing with non-normal distributions, including fat-tailed distributions.  So Talib is just wrong.  He didn't say normal distributions were more tractable, or easier to deal with, or cet. par. require less data to reach significant conclusions, all which would be justifiable.  He said statistical tools "do not work," no qualification.

The key bit is "Their statistical tools".  He's assuming (correctly in many situations) that they aren't using the tools that can deal with some non-normal distributions.

That was my interpretation as well, and I think Joan is reaching a little in his interpretation.  If Taleb believes that there are no statistical tools to deal with fat tails, then my guess would be that he wouldn't publish a paper where he uses statistical tools that deal with fat tails.

Well we can't really judge his interpretation easily because there's no literature review in the paper, and neither has the paper passed peer review yet.

EDIT: I sit on the editorial board of a journal - they wouldn't be the first two well known academics in their field to make grand pronouncements about the methodological rigour of another field and be completely wrong.

[The Minsky Moment]

The key bit is "Their statistical tools".  He's assuming (correctly in many situations) that they aren't using the tools that can deal with some non-normal distributions.

It's a pretty simple set of 2 English sentences.

"But social scientists do not get it. Their statistical tools do not work for what we call fat tails."

There is no qualification, no nuance, no explanation.  He is singling out the entire category of social scientists and saying their methods don't work. Period.

[The Minsky Moment]

If Taleb believes that there are no statistical tools to deal with fat tails, then my guess would be that he wouldn't publish a paper where he uses statistical tools that deal with fat tails.

Ah but that's the point isn't it?  Taleb is the great iconoclastic genius who alone truly understands, who will save the masses from those misguided social scientists who can't grasp his brilliant insights.  (unless they buy his most recent book . .  on sale now!)

I easily lose patience with this kind of nonsense.

[DGuller]

If Taleb believes that there are no statistical tools to deal with fat tails, then my guess would be that he wouldn't publish a paper where he uses statistical tools that deal with fat tails.

Ah but that's the point isn't it?  Taleb is the great iconoclastic genius who alone truly understands, who will save the masses from those misguided social scientists who can't grasp his brilliant insights.  (unless they buy his most recent book . .  on sale now!)

I easily lose patience with this kind of nonsense.

Sounds like you have prior issues with Taleb.

[The Minsky Moment]

Don't know him, never met him, just read his stuff.

[DGuller]

I'm embarrassed to admit it, but I've never read Fooled by Randomness.  This thread finally made me go and order the book on Amazon, to see what he himself has to say.

[DGuller]

That said, from what I know of the main argument that he has to make, his point is definitely one that needs to be made much more often.  Maybe he's not the first one to realize that ignoring those rare outliers can completely invalidate your sophisticated-looking models, and embark you on a dangerous path as you make decisions with it, but he's definitely the first one to make the masses think about it.

[The Minsky Moment]

DG I don't have a problem with that.  I think popularization of ideas like that is a very good thing.  But it ends up being counterproductive when it involves dubious sweeping unqualified claims, accuses everyone else in his field of being fools, and attacks everyone else's motivations. 

Look at someone like Daniel Kahneman - that's a real bona fide genius, who really did come up with theoretical insights that demonstrated fundamental flaws in basic methods of a number of fields, did vast amounts of empirical work to verify and hone his theories, AND popularized it to boot.  And yet managed not to be a arrogant prick about it.

[Admiral Yi]

I thought it was a trivial point Joan.  The types of academics who look at things like declining warfullness trends don't usually get very far into statistics.