If you're not the customer...

I'm guessing someone (or something) made the argument that forcing people to scroll through every video uploaded to a channel just to get to the oldest ones "increased engagement"

If it's anything like redesign work I'm used to, someone redesigned and just happened to remove it with little reason for it. Then the numbers went up as people had to scroll so they assumed everyone loved the redesign and didn't question why.

I think this hints at a concept I find very hard to articulate about letting metrics guide your design. You set out with metrics that seem plausible, you work to improve them, and if they do improve, you call the design a success. But then it actually turns out the metrics were no good in the first place because they “improve” for scenarios where you’ve most certainly made the user experience worse. Metrics proponents would argue that you should have had better metrics, but then I start to feel that the problem of selecting the right metrics becomes an art form just as subjective and complicated as designing a good UI in the first place, at which point they’re no help.

Goodheart’s law may not precisely capture what you’re getting at, but seems closely related

https://en.m.wikipedia.org/wiki/Goodhart%27s_law

Engagement is a nuanced metric for that reason. Am I engaged because I am enjoying the product or is it just taking me longer to achieve things because the UX is bad.

> selecting the right metrics becomes an art form just as subjective and complicated as designing a good UI in the first place

this is absolutely the truth!

I would say that perhaps what is lacking is a clear vision of the product. Steve jobs didn't use metrics for the initial design of the iphone.

I suspect that as a company grows bigger and bigger, the product loses the initial visionary that created the success in the first place. May be the "founder" (of the product, not necessarily of the actual company) left, or moved on, and what remains is just diluted responsibility between a few different people. Therefore, they resort to metrics, as they have no idea what works, nor have a vision.

Knowing Google, it was probably a feature that someone didn’t think was important enough in a UI refresh to spend time on and was added now because the team happened to have time (or the interns have started).

That wouldn't make sense. YouTube wants you to watch ads, not click around the UX.

Time spent scrolling is not time spent watching ads. So that is definitely not a metric YouTube is optimizing for.

I have no doubt that total time spent in the app is a metric they track. Plus scrolling through all of those videos on the way to the video you're actually looking for means there's a decent chance you'll see a thumbnail or title that catches your interest leading to more time watching videos. Either way, It'd be a shitty move and one that somebody should have caught and corrected a long long time ago.

I imagine the cost of the index used to store videos by oldest uploaded outweighed the likely marginal number of users who used it?

I think inexperienced developers fall into the trap of implementing whatever they are told to, probably to show that they can. I think it's important that developers think about what they're being requested to do, and suggest alternatives if it makes the software crappier from the user's perspective.

Do you think an inexperienced developer at YT made this change?

Maybe? Generally bad UX decisions are opposed by more experienced devs, often because of the edge cases that are not considered by designers, while less experienced devs just do it.

My guess is that it was a feature that was used rarely and only by a tiny proportion of the userbase. That kind of feature is always at a high risk of being cut if it causes even the tiniest problem. In this specific case, if an option that's used 0.01% of the time is making the UI look cramped when the contents of a dropdown get inlined, why not remove it? Nobody's going to care.

Except that sometimes people do care. It's pretty hard to know in advance whether https://xkcd.com/1172/ will apply or not. You can't run any kind of UX study because odds are that the only way you get people to interact with the feature is by making up an artificial scenario that basically forces them to use it. You can't even run an A/B test because every fewer people will actually reduce their YouTube usage over this. They might be unhappy, but their level of happiness doesn't show up in metrics.

I suspect that YouTube would much prefer to move away from being considered a video repository, and instead be considered a video feed.

I don't know the figure, but if I had to guess, I would wager that 95% of their views at any given time go to videos produced in the most recent month. Video is incredibly expensive to store, and almost all of what they're storing is basically sitting in a graveyard.

based on experience there could be a bunch of pretty innocuous reasons e.g. 1. some company wide simplification effort on user options 2. some weird coupling of the UI project to the backend project, where this functionality was an edge case that they could postpone til after launch

"Postpone until after launch" === "Never build"

Perhaps not in this case, since it's now back.

The answer is money. It's always money. I absolutely 100% guarantee they ran some kind of analysis and found they made less money when people looked at older content per channel, on average.

I always wonder about YouTube’s storage infrastructure and whether they move old videos to cold storage in a way that is costly to recover, and that maybe that’s why they try to discourage watching older videos… but that is probably a bit oversimplified.

Anecdotally, there must be some difference, as it often takes longer to get a video started if it's old.

This has always been my suspicion, considering older videos (or just videos with not a lot of views in general) tend to take noticeably longer to load.

My immediate assumption is these videos are not edge cached, so naturally they want their platform to avoid delivering these types of videos. They would rather stick to the well-cached, proven popular videos that are currently circulating.

You can’t store old content in long term archives with slow access rates if people keep sorting by old just to randomly view things. It’s terrible for caching