Will Google Kill Science?

Not an obvious conclusion, I'll agree. However, Chris Anderson, editor of Wired, presents the argument like this: as all sorts of data accumulate into a vast ocean of petabytes, our ability to synthesize it all into elegant theories and laws will disappear. The story is the cover of this month's issue of Wired but I came across it in a newsletter from The Edge, a group of thinkers trying to promote a "third culture" of online intellectual thought.

Anderson's argument isn't really that the scientific method will disappear, but rather that correlation will become as good as it gets in terms of analyzing real-world data. Everything will be too messy, noisy and changing too quickly for proper hypotheses and theorems. As Anderson puts it, it will be "the end of theory."

The nice thing about reading this on Edge is that the newsletter comes with several critical responses included from "The Reality Club," which includes thinkers like George Dyson, Kevin Kelly and Stuart Brand. But I say that as the consumers and producers of most of these masses of data, the vote should lie with you, reader: does Google's brute force approach to data hording spell the end scientific elegance?


by | Permalink | Comments (2) |





Comments (2)
You can skip to the end and leave a response. Pinging is currently not allowed.
  • Pete says:

    Chris Anderson is wrong about the end of models. Models predict and extend. Even given complete historical and current data, there is nothing predictive about it. In science as in finance, past performance is no guarantee of future results, and just as a crowd behaves differently from a group of individuals, two sub-critical lumps of uranium behave differently separately than they do joined together. Even if you take complete information about what happened yesterday and today and try to extrapolate tomorrow, you are using a model: you are assuming that tomorrow will look something like yesterday and today.

    Just as a recent Harvard study showed Chris Anderson was wrong about the long tail, he is wrong about correlation and models. However, popular culture will increasingly democratize knowledge — the more people believe something, the more relevant that belief will be. Even if it is wrong. This is why Barack Obama is a crypto-Muslim radical Black Christian.

  • Ed Finn says:

    Pete,

    I agree with you that Anderson is wrong and the era of models is certainly not over (just beginning, actually). The point you make about Obama is really in Anderson’s favor: with so much information out there, many people have given up hope of reliable, objective truths and just put faith in facts the Google way–based on the number of times people repeat them. Maybe that’s always how we did it, and there are just more voices out there now.

    Actually, I think your comment touches on a separate problem. Science will do fine, and the masses of data will be analyzed and understood by an ever-more sophisticated apparatus of statistical tools and algorithms. The trouble is, the math necessary to cope with this brave new world is not being taught to enough of the people who need to understand it. I can’t tell you how many times as a journalist I saw blatant misrepresentations or misunderstandings of correlation and causality. As the data and the science get more complicated, the Average Joe and even the Average Joe Journalist is falling behind.

    Thanks for your input!
    Ed

Leave a Reply

Quantcast