The Malcolm Gladwell Problem

I’m not the first person to say this, but I’m increasingly vexed by Malcolm “Tipping Point” Gladwell’s stuff. Case in point: MC’s piece in the current issue of the New Yorker on neural networks for predicting movie and music hits. All of Gladwell’s tics are on display in service of an article that mostly feels bottom-drawer, like something rejected twice and then run out of sympathy.

An example: Gladwell cheerfully quotes Platinum Blue CEO (Mike McCready) on the wonderful job his software does separating hits from flops in the music business. His evidence? McCready claims to have ID-ed Norah Jones “Come Away with Me” album back in 2002 as a monster. How do we know McCready did this? He says he did, and Gladwell quotes him saying “a local newspaper [in Barcelona]” interviewed him then and he told them. But we are not told which newspaper, which day, etc. And we’re not told how many false positives McCready has had, where he told someone something would be a hit and it wasn’t. I’m not saying it’s impossible; I’m just saying Gladwell has done nothing to convince we readers that we are to take this stuff seriously, other than offer his say-so.

There is more of this later in the article, with Gladwell somehow engaged with a mysterious bunch of pseudonym-using, software-toting movie auteurs. They say they have a way to say in advance how much movies will make, and how to engineer them for better performance. This time we have some real data, which goes like this:

  • Nine movies analyzed
  • On three movies the software bombed in estimating the final gross
  • On one of remaining six the software called for a gross of $49m and movie made less than $40m, but Gladwell calls it a success
  • “On a number of films, they were surprisingly close”, but we’re not told what close is, other than that it was within a few million on at least one

So, let’s charitably say that the software did decently on 3 out of the nine movies it looked at. Is that really an improvement on the status quo, especially when you don’t know, a priori, which three? I don’t think so.

Update: Added link to Gladwell’s piece, which the New Yorker has now made available.


  1. silvester says:

    I think by a priori you mean ex ante.

  2. Gladwell should write an article on Complexity Theory and the Santa Fe institute–how it sounded like the most brilliant thing since sliced bread when introduced, still continues to fascinate in an intellectual sense, and yet falls far short of its promise (so far).
    Then he should turn the piece into a self-critique of Gladwell as meteor, segue into a brilliant discourse on irony, and sign off with news of shaving his head / embracing existentialism / becoming a philosopher poet.

  3. I only saw the video snippet on Norah Jones. He asked, how could someone discover her? I wonder if MG ever heard of Ravi Shankar – It’d be like, how would someone have found Barry Bonds, or Ken Griffey Jr. – their dads were stars also. He should have at least used a different example.