Hacker News Reader

It's actually sequential testing which is not necessarily Bayesian. Also, it's not a cheap knock off. Just another way of getting to an answer.

View on HN

Bayesian updating of the posterior? Or if you prefer frequentist algorithms for online learning of classifiers?

Not trying to pick a fight, just as a statistitian/ML developer I've seen the same things be reinvented and renamed so many times.

No, if you read their technical paper, it's frequentist sequential testing with false discovery rate control, which is a fairly recent development (I mean, 25 years old is pretty new in statistics).

http://pages.optimizely.com/rs/optimizely/images/stats_engin...

I think all OP is trying to point out is that it either agrees with bayesian methods or it's wrong... so at best it's not materially new, and at worst it's using questionable assumptions.

"it either agrees with bayesian methods or it's wrong"

This kind of faith-based statistics is pure ideology.

Reply on news.ycombinator.com