[kccqzy]

I don't want to speculate baselessly but if you've worked in big companies you'll know seemingly simple features become complicated with scale. This is especially when you are YouTube. It could be that the old feature was causing production problems and had to be removed, while a more efficient re-implementation is underway. I mean just look at what Google's Cloud Spanner says about using timestamps as keys: https://cloud.google.com/spanner/docs/schema-design Having both sort by latest and sort by oldest in an efficient way could very well be a nontrivial problem.

[thrdbndndn]

I agree with you in general, but in this particular case? No, I really don't believe it's a technical issue.

It of course may not be as simple as your random hobby projects, but Google has the manpower that is proportional to the complexity of YouTube.

[mardifoufs]

Isn't that the type of sorting you basically only need to do once? Or I guess in the rare cases when a video gets taken down, privated or relisted? I know it's not super rare but it would still be a much rarer event than users sorting by old. So they can just keep the list or set of video IDs, no need for expensive db operations right?

(I'm being a complete armchair expert here, since I have no idea of the actual state of the YouTube backend. It might look simple to me but I'm sure it's not in reality . I'm curious to see what would be the best way to optimize a feature like this one!)