Hacker News Reader

I am a little sceptical of using A/B tests on accessibility features. What, precisely, is being measured here?

The acceptance criteria for accessibility features are narrow to a specific group of users (actually, the users fall on a continuous spectrum). Furthermore, there are legal, ethical and moral obligations. Would those be thrown out if the A/B test was a fail?

View on HN

> Would those be thrown out if the A/B test was a fail?

It sounds like it, yes.

> there are legal, ethical and moral obligations.

From the evidence in this article, Trivago don't give a shit (or didn't)

In these cases, no, they wouldn't be thrown out. The aim is always to accept accessibility changes but they still need to be tested first, just in case they have negative affects which would need further investigation to be fixed.

Reply on news.ycombinator.com