If you rely too much on A/B testing it is easy though to get stuck in a local minimum defined by users' expectations... Or to build a site that generates a lot of revenue but that's not optimal for users, and when a competitor breaks out of the local minimum by changing the paradigm, all the users go there.
This is true in general, but in this specific case looking at the biggest online marketplaces (e.g. amazon, aliexpress, ...) they come from completely different places and have converged to the same kind of conclusion.
There might be other paradigms looming out there, but the current situation is not just people lazily relying on A/B testing.
The same kind of conclusion? Amazon and AliExpress have wildly different design philosophies. One has pointless, hard to use coupons and mini games, and the other is Amazon.
I use AliExpress _in spite of_ their UI, not because of it.
Aren’t both at the same level ? Amazon is horrible in many many ways.
You’d think a bookstore would have a good way to show books of the same series in order, and make it easy to find any specific issue. Or help you follow random authors, and not just some handpicked ones, or notify you of a new book coming out in the future. All of these use cases are mildly (wildly) broken and it was the core of their business for a long time.
When I say they are similar, I’m thinking mostly about how they’re ignoring “pretty” and “logical” design trends and stick to very basic while chaotic page designs.
It's just as often that one optimizes for minima than for maxima (eg. latency or revenue). When there isn't a natural orden it's a coin flip (eg compounding latency and revenue in a single function). And anyways both are functionally equivalent by means of order invertion.
Personally, when speaking in abstract I stick to saying "local extremum" just to avoid this nagging question.
