Gladwell and eigenvectors - Stephen Baker

Home - Viewing one post

Gladwell and eigenvectors posted on November 25, 2009

General

I'm only catching up now to Steven Pinker's skewering of Malcolm Gladwell in a New York Times book review. (What is it these best-selling authors have against each other? A few months back, Gladwell took Chris Anderson to task.) Pinker criticizes Gladwell for his lack of "technical grounding." As I read it, I thought, "Jeez, if only my book had been the mega best-seller I'd dreamed of, maybe Pinker (and others) would be tarring and feathering me." Yes, it's healthy to remind myself, there are advantages to not being Malcolm Gladwell.

Here's Pinker:

An eclectic essayist is necessarily a dilettante, which is not in itself a bad thing. But Gladwell frequently holds forth about statistics and psychology, and his lack of technical grounding in these subjects can be jarring. He provides misleading definitions of “homology,” “sagittal plane” and “power law” and quotes an expert speaking about an “igon value” (that’s eigenvalue, a basic concept in linear algebra). In the spirit of Gladwell, who likes to give portentous names to his aperçus, I will call this the Igon Value Problem: when a writer’s education on a topic consists in interviewing an expert, he is apt to offer generalizations that are banal, obtuse or flat wrong.

I feel for Gladwell here, because in researching my book I was often immersed in sciences and formulas I understood only through analogy. In fact, the Numerati would occasionally use me as a courier, carrying messages I didn't understand from one to the other. One of them involved the very term Gladwell flubbed, Eigenvalues (or their first cousin, eigenvectors).

I was at IBM Research talking with William Pulleyblank. I told him I'd be seeing his former colleague, Prabhakar Raghavan, the next week at Yahoo Research. He told me to ask Raghavan the following question: "Aren't you effectively computing the eigenvectors of a symmetric matrix?"

Duh, I said. OK. So I carried the message across the continent it to Raghavan. Here's how he answered:

"There's this recurrent them of eigenvectors that came up in link analysis, and page rank. You can think of that as propogating trust. But curiously, it says nothing about propogating distrust. it's a curious thing mathematically. The trust propoagation page rank can be viewed in terms of propabilities....The propagation of distrust is like having negative probability, and there's no such thing, at least in the mathematics we understand. This is why it requires new inventions... we don't understand the stuff..."

Now, I don't know how many of you understand that. I get it conceptually, but not mathematically. So if I decide to write about it, I cling to the concepts and tread very very carefully when it comes to the math. It's limiting. But because I'm not Malcolm Gladwell, people like Pinker rarely take me to task.

2 comments [view]

add comment

send to friend

Cog psych yesterdy at Penn St. Try counting things w/out moving finger. You rock, nod, or tap foot,anything to create rhythm.

follow me on twitter

The Book Bag - Zoe Page

The Wall Street Journal - John Derbyshire

Frankfurter Allgemeine Zeitung - Milos Vec

The Guardian (UK) - Steven Poole & Christopher Exeter