Taleb’s new big idea

I have rarely looked forward to a book as much as I looked forward to The Black Swan, which may be why I found it so disappointing. Taleb deserves a lot of credit for his strong stand against the presumptions of so-called risk management in finance, and for his efforts to bring a proper understanding of probability and uncertainty (another way of saying, a deep sense of skepticism about economic models) into discussions of markets and the economy. But his unending fascination with his own life story as well as a bizarre repetitiveness made his first two books – particularly the second – nearly unreadable.

With appropriately guarded expectations, I was interested to see that Taleb is coming out with a new book called Antifragile. The Farnam Street blog (h/t Ritholtz) recently posted a brief excerpt from the book focusing on the balance between signal (useful information) and noise (everything else). The excerpt hints at Taleb’s larger argument, which appears to be that having too much data creates excessive noise, which in turn creates a temptation toward excessive intervention with poor results (typically for Taleb, this is not described simply as “unintended consequences” but in the more showily abstruse medical metaphor of “iatrogenics”).

I look forward to reading the book – information is a fascination of mine at the moment, and Taleb is always at his most interesting when he is onto something new. But even before reading it I am left wondering how he will address the central problem of managing information overload – the fact that the selection of the filtering mechanism is among the most political of acts, and is one with profound consequences.  In the excerpt, Taleb seems to hold out Greenspan as an example of someone who drowned in excessive data. But the bigger problem with Greenspan was surely the opposite – that he filtered all of that data through a set of beliefs so rigid that they could only allow Randian conclusions. And looking more broadly, isn’t this sort of filtering a major problem in American political culture at the moment? The proliferation of media allows us to allocate our attention to information sources that more or less confirm what we already believe. Far from leading to a society that is better able to manage and process information, filtering seems to be having the opposite effect.

Even in cases where the selection of the mechanism is not politically driven, it still has profound implications that can often obscure more than is revealed, a characteristic inherent in the mechanically statistical approaches to risk management that Taleb has rightly derided. The most obvious example is the use of value-at-risk (VAR) by Wall Street banks and their regulators as a means of distilling multiple dimensions of risk across countless trading positions into an easily digestible, single number. By adopting this practice, banks have arguably made themselves even more fragile by their over-reliance (if only for show) on a measure backed by the “science” of financial economics. At that point, sociology (the reflexive structuring of organizational behavior around a practice) and psychology (narrow framing, etc.) kick in, and you end up with a powerful filter of unruly data that leaves management even less informed than they were without it.

I have to assume that this is what he is referring to in the title of the book but it’s not clear that this is what he means from this admittedly brief excerpt. I have preordered the book and look forward to seeing where he goes with it.

Advertisements
This entry was posted in Information and systems. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s