I realized with some sheepishness, after writing about Bruce Sterling's story "The Black Swan," that I had only read second-hand accounts of Nassim Nicholas Taleb's Black Swan idea, but not his actual book of that title. So I ordered it, and now have read the book itself. A "black swan" as defined by Taleb, is an event of "rarity, extreme impact, and retrospective (though not prospective) predictability." Taleb is an ex-quant turned iconoclastic mathematician. This book was a bestseller, and like many best-sellers on arcane topics, it is chatty, opinionated, and a bit on the loose and sloppy side.
As I expected, I enjoyed the book a lot. His subject matter appeals both to my taste in mathematics and my aesthetic of fiction. (My favorite section of David Hartwell's The Dark Descent is The Fabulous Formless Darkenss, which is the section for what he and I call "nature of reality" horror -- stories in which the parameters of what one thought was possible usually change for the worse.)
Other than having it confirm my belief system (always a pleasure), there were two basic things I got out of reading this.
As I told my husband (David Hartwell) while reading the book, for me this is partly a book about how David's way of doing things is right. The way David publishes books as an editor, the way he runs our small press, the way he collects books, the way he runs the family finances, etc. all tend to limit downside risks while at the same time leaving him exposed to positive Black Swans -- serendipity. (Aren't I a good wife for blogging about how I just read a book that tells me how right my husband is?)
In the second half of the book Taleb spends a lot of wordage and emotion on condemning the "intellectual fraud" of the Bell Curve. I don't really need to be able to judge whether he is correct in terms of all of his examples. But it set me thinking about standardized testing.
Last week and this week, the school district is doing extensive testing of my son, which they do every three years. (This is separate from the group standardized testing.) When my husband I meet with the district in June, we will be presented with a sheaf of paper in which, page after page, our son will be raked in terms of percentiles and numbers of standard deviations from the norm. And we will try to make sense of all this data. Again.
Years ago, I read Stephen J. Gould's The Mismeasure of Man (another book which confirmed my belief systems) and I have a deep distrust of standardized testing anyway. And yet these tests mean something, and it is of crucial importance that I understand what. In past cycles of this, I've wasted a lot of time reading up on what is meant by "processing speed" and how to understand things like the results on subtests of IQ tests.
My son tends to produce an anomalous pattern of results on such tests, and so far no one has yet to be able to explain what this actually means, though -- who knows -- this year's tests may be different. (We expect to be having our independent expert read the results for us this time.)
So I'm reading along through Taleb's attack on the uses of the Bell Curve, and I realize that of course all of these tests are normed to Bell Curves. They wouldn't give them if they weren't. But all around me I see human attributes that do not fit that model: the ability to draw a recognizable portrait, the ability to play the piano, gender, eye color, etc. Many human differences are not meaningfully normalized to the Bell Curve.
And so I realized there was one further aspect of what these tests measure that I need to keep in mind: that the imposition of percentiles and standard deviations on what attributes are measured may be no more realistic than contemplating a spherical cow. We somehow need to try to consider whatever we're told each of these these tests are measuring is distributed Bell-Curve-like over the population, or whether it is a different kind of thing. I find this a bit daunting, though probably worth trying.