Big data, small data, crap data

Big data.

“If you’re not into big data now, then you will be soon” is often heard – usually at IT and governance conferences. And itt might be all rubbish!

Measuring stuff in millions of different ways is becoming the norm without really understanding why.

Conventional wisdom goes something alone the lines of…

  • We break what we do into chunks and call them objects
  • Objects have properties hence objects are a collection of properties
  • Measuring the properties allows us to understand the objects
  • Affecting the properties then changes the value of the objects

Except that’s assuming a number large number of leaps of faith. Properties do not define the usefulness of an object: The processor speed in a mobile phone might be irrelevant and probably not as important as it’s battery life.

Creamy bikers get creamed

At it’s extreme, the correlation between unrelated items becomes pivotal in created terrible decisions. For example, the per-capita consumption of sour cream had a clear correlation with the number of motorcyclist killed in non-collision accidents.

Not everyone is suckered into to measuring and correlating items, but it’s easy to do. Try and avoid it next time you’re asked about “bounce-rates”, “pageviews” and the like. If they translate into revenue, or cost savings, or service then by all means.

If not. Just say no.