[ricardobeat]

You just measure user behaviour to prove your hypothesis. If your assumptions were wrong, you’ll get null results and move on.

An article like that coming out would bias and invalidate the test.