The ineluctable limits of logic

February 20th, 2013 by Jay Shaw Leave a reply »

David Brooks just wrote an opinion piece in The New York Times, What Data Can’t Do.

It’s really good; reasoned, nuanced, supported by relevant example and contextualized in the literature. Well worth reading.

It got me thinking about the limits of data-driven decision making in general. Bertrand Russell argued that no system can be understood by reference to anything from inside the system, no matter how clever the arguments. Ludwig Wittgenstein argued that nothing can be understood, or even adequately explained — period.

The data scientists are giving us amazing insights and trying to explain everything. They will doubtless continue to generate lots of actionable insight, including insight into the world of talent management, but they will fail at the “everything” part.

Everything is too tall an order.

Read Brooks’ piece. It’s food for thought.

3 comments

  1. Coen Flach says:

    Jay,

    cannot resist pointing to an article about the difference between correlation and causation called “Storks deliver babies”:

    http://www.uni-marburg.de/fb21/motologie/mitarbeiter_seiten/ls/storks.pdf

    I will read the article first though. Thanks for pointing it out.

    Coen

  2. Jay Shaw says:

    Coen — you’ve ruined my day. I was sure storks delivered all of the babies, no exceptions. Now it looks like the statistical evidence may be less than perfect. The correlations, though high, are not exactly one to one.

    Time to bring in more data analytics firepower I guess . . .

  3. Klaus Loeper says:

    I now have read both of the articles and am sitting here literally scratching my head.

    Hmnnn.

    What about some more statistics?

    http://www.lifeslittlemysteries.com/2057-8-weird-statistics.html