| First, I really wish we would stop calling random stuff we don't like (even for good reasons) evil. It blurs the discourse and hampers objective discussions IMO. I understand that A/B testing can be used to exploit addictions mechanisms and other gray (at best, more likely dark) patterns and IMO that should be forbidden. At the same time, I don't know any mobile company launching a new feature without a server side A/B lock these days. Simply because that way : - it allows you to open the feature gradually and close it if there is a nasty side effect like a crash. - look at usage patterns between 2 features. I guess the separation between helping the user and exploiting them can get blurry but really when we look at data it is to see if they know the feature exists and find it useful. The truth is that almost all users barely know how to use their phones, so each time there is a feature based on long click, gesture, something buried in a screen other than the first one of the app, a button in the toolbar, etc, chances are that most users won't even know that it is there, even if they would like to use this feature. I honestly don't know how to conciliate the 2. |