PSA: Watch in an private/incognito tab/window. If you are currently logged into your google account, this WILL pollute your watched history: https://www.youtube.com/feed/history
You can click X on them and quickly remove them. If there's truly way too many (you left it open for hours), you can go to your Google's Activity page, filter youtube and delete everything from today with a button.
When I go to YouTube on a fresh device without being logged in, it's a pile of steaming clickbait and pop-internet-culture garbage. On my account, by contrast, YouTube is full of mostly great recommendations of high-quality content and I can pretty much always find several new and interesting things to watch should I feel like it.
(I watch stuff like Kurzgesagt, Smarter every day, AvE, Rick Beato, Today I found out, Wisecrack, Wendover Productions, Practical Engineering, Vox, Crash Course, SciShow... that's just from browsing my current recommendations. I would guess none of that shows up for fresh accounts)
The problem I have is that my recommendations are usually full of stuff I'm already subscribed to. Discovering new things is really hard and usually just happens by reading about it elsewhere.
Clean it of trash and you will receive less trash recommendations. Fully cleaning isn't effective as the average video quality is lower than most HN public would like to watch. Having mostly good videos you like makes it recommend more similar content to you.
It's not just average video quality, it's that (as far as anyone can tell), the algorithm optimizes for expected value of total watch time, as opposed to optimizing for just the likelihood of watching the next video. Basically, the recommendations behave as though they "hope" that you will go down a conspiracy or outrage rabbit hole and binge 12 hours of garbage.
For me it correctly "hopes" I'll go down science rabbit holes, which I do. I'm similar to the person above, I watch a lot of educational content, and I also get suggested great content.
Right now I see the follow up to the guy who build his own VGA card, video of someone building a camera that can see wifi, and a PBS video on the "quantum internet". All great suggestions.
Also consider using Firefox and enabling `privacy.firstparty.isolate`, which will separate the cookies for third-party embeds from their own domain, thereby preventing this (as the embed doesn’t see you as logged-in).
they probably log a 'shadow view' but recommendations seem to be directly based on your watch history since your recommendations are lost when you completely clear it.
The recommendations will mostly be derived from the videos you have saved in playlists. And the front page will always shows Youtube Mixes about the last song you were listening to, at the time of turning off the watch history.
Do a significant amt of YouTube users use playlists? I’ve always assumed no. Unless perhaps they are doing it for music since you’re talking about YouTube Mix. And songs.
Some like me actually appreciate the youtube recomendations based on the previously watched content, but need to be extra careful when watching some kind of content that is likely to be weighted a lot by "power users". Example, I do not follow videogames, but I do enjoy watching speedruns of old games ocassionally, so I need to watch it on incognito mode so I won't have my recommendations flooded with videogame videos.
I'm currently unaware of any recommendation engine that's worth acknowledging.
Netflix, Amazon, Youtube, PornHub, etc... they're all accomplishing little more than "similar to the one, and only one item you last saw", with dramatic shifts in "profiling" from one or two videos.
Actually, Netflix acknowledges this and splits the recommendation into "because you watched X..", so at least it covers a greater range (eg last 5 things seen)
I'm damned sure they could be much more useful if they would let me tell them what I like, by implementing rating systems that are worth using (e.g. the ability to browse and edit previous ratings in a sane fashion)
but user-useful recommendation is not the actual goal, so really its just that our metrics are wrong. It's probably great according to view counts.
That's the bottom line right there: thinking that recommendations are useful to YOU is naive and missing fundamental things about the experience.
That said, you can indeed tell them what you like but by use of negative space. When you get bombarded with obviously horrible recommendations, do the two-step process of clicking 'Not Interested' (if possible without even watching the video, or you can check it in incognito mode, assuming they're not watching that even more closely) and then 'tell us why', and respond 'I'm not interested in this channel: "undesirable video maker".
That assumes you can be sure you want to nuke the channels and subjects in question, but when it's clickbait channels and/or alt-right propaganda it's generally easy to identify and not get wrong. I'm sure the same would be true for leftwing propaganda, but the stuff I don't want pushed on me has a whole language and lexicon that's easily recognizable by video title, channel title and attempted clickbait image. If stuff trips my sensors on those grounds, I'm generally comfortable nuking it unseen.
The only one I think is good is when I build radio stations off custom playlists on Google play music. Generally speaking, most of the songs are good, and like 40% will get added to the playlist as well.
I don’t use this, or Spotify, but those are the only two systems I’ve heard people give praise for the recommendation engine — I suspect that its because music playlists are almost equivalent to a rating system.
That is, the user is capable of efficiently informing the engine of their taste, and there’s significant incentive for the user to consistently re-evaluate their ratings (playlists), so it can be trusted as up to date.
Another very important aspect is that playlists are useful enough to the user that they actually want to maintain it.
For example, amazon, netflix and pornhub all have rating systems, but they’re not at all useful. The interface isn’t useful enough for reviewing and reflecting on, its not comprehensive enough to keep as a primary list (because it only covers what they offer, which is very limited) and there’s of course no impact on the recommendation engine (because the rating systems are not worth using; chicken and egg). No sane person would touch the things (beyond “upvoting”, which isn’t significantly related to taste)
Imo ratings are absolutely vital to useful reccomendation, but they’ve been totally neglected
The radio stations off custom playlists are indeed pretty good, but the recommended ones on the home page are often horrendous and very short (songs will basically get repeated after less than 1 hour).
last.fm is great for recommendations after you've been scrobbling for awhile. I've picked up so many artists from checking out recommended every once in awhile.
On this subject, clicking through to some of these videos in incognito gave me what I guess must be the "default" suggestions - which included not just beauty tips spam but far-right speakers and loyalist marches in Belfast. I'm guessing the latter is related to the date but it's not something I'd expect to be the default! Watching even one video then made it more normal, even when trying to generate recommendations based on the no view videos.
Or otherwise paid for (which I think still counts as 'ads'), or tuned towards engagement.
I think quite a lot has to do with machine learning working out that if you panic the human animal they pay attention to threats, and therefore to maximize engagement 'if it bleeds, it leads' (old newspaper maxim). Newspaper editors can (and might not) automatically apply a social-benefit heuristic or sense of social shame (not wanting to be a 'muckraker' or troublemaker), and machine learning may not even start with such a concept.
If engagement was maximized by turning viewers into cannibalistic humanoid underground dwellers (CHUDs), machine learning would simply make note of that and run as hard as it could in that direction, since it doesn't have a larger context in mind unless programmed to do so.
(Such a larger context is actually sort of controversial: lot of people demonize the very concept of social justice, and without it you get these hacks to maximize engagement by tapping into really unmanageable human/animal behaviors)
Because a lot of people care of recommendations? If you watch news on yt then maybe recommendations are garbage but if you watch videos related to some niche activity or music then recommendations are a godsend.
Oh, you think you can actually disable it? its just an UI setting, Google keeps tracking you anyway. Go ahead and try it, disabling does nothing to the recommendations.
Others are suggesting practical reasons for having watch history enabled, but I would think that the vast majority have it enabled simply because that is the default.
history augments memory when searching for specific things one remembers. having it polluted by random videos embedded on third party might not be great. history also drives recommendations, so again having random video views will ruin coherence of the home page suggestions
I wish you hadn't told me that. I thought I saw something suspicious when I resumed switching and yep, that's a bunch of topless elementary schoolers at a pool party. I suppose it doesn't technically violate YouTube's policies but jesus I did not want to see that.
PSA: Watch in an private/incognito tab/window. If you are currently logged into your google account, this WILL pollute your watched history: https://www.youtube.com/feed/history