Hacker News Reader

I believe this applies mostly to public science. Private science is anchored by a market-based mechanism that naturally expunges fraud.

Is this a joke? Profit motive is far more likely to put the cart before the horse: pushing for evidence to support outcomes.

I think they were referring to internal research that says 'if we do X we can sell more' which has naturally aligned incentives.

This as opposed to releasing research along the lines of 'cigarettes are fine for you! smoke more!'

> if we do X we can sell more' which has naturally aligned incentives.

Even then there are agency problems which mean that once a company gets over a certain size there is no natural alignment.

I know plenty of managers tweaking A/B tests until they get the results they want.

Easy fix: hire a statistician. :-)

"in academia overfitting can get you a nature paper, in industry get you fired"

Not so much. When you pay someone to get a result, that’s what they do.