Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Mozilla Is Crowdsourcing Research into YouTube Recommendations

foundation.mozilla.org

Post reply on HN
Print view View original

151–160 of 179 posts

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#151

Post by anjc »

srean wrote: post #26
FiReaNG3L wrote: post #5

What's impressive to me is that YouTube Recommendation are so ... stale. All it recommends to me is content from the same 3 categories, and the same videos over and over and over (you would think if I didn't click on the suggestion in the past 6 months it would adjust). Now the topics it recommends to me are non-political, but I can very well see that if you fall down one of more extremist rabbit holes thats all you…

For me it has been the opposite as in it has so little hysteresis. Just one day I watch something I do not usually watch, say cat videos, and for weeks I will get only cat videos as recommendations. I think their algorithm needs to be cognizant of and adapt to temporal shifts in a viewers interest. Some users may like change in focus at slightest of deviations, others may prefer more hysteresis.

This criticism shouldn't apply anymore. They predict categories you like and you can select a category. The 'All' page should be a mix of all categories, not just cat videos. Amazon on the other hand...

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#152

Post by anjc »

thesuitonym wrote: post #118
anjc wrote: post #63

I'm always mind-blown reading comment threads like this regarding recommendation, with people annoyed that explicit feedback helps recommenders, or exasperated that they're recommended one item that they haven't approved while also wanting novel recommendations. 1000s of hours of videos are uploaded every minute. 5+ billion videos are watched daily. Are you not rather quite pleased that YouTube can deliver highly rel…

Consider that a lot of the complains are that YouTube constantly recommends already watched videos, instead of new and relevant ones. Then, if a user watches a single video on a topic outside of their usual interest scope, YouTube replaces any semi-relevant recommendations with with that new category, regardless of if the user is actually interested in the topic.

> Then, if a user watches a single video on a topic outside of their usual interest scope, YouTube replaces any semi-relevant recommendations with with that new category, regardless of if the user is actually interested in the topic.

That's the way it used to work, now it adds that new category to the list of categories it thinks you like (top of the home page), and you can filter or include that category.

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#153

Post by throwaway316943 »

anjc wrote: post #87
throwaway316943 wrote: post #82

Earlier quoted context omitted.

The complaints are due to a significant decline in the quality of recommendations. It’s like that scene from 2001 where HAL is slowly lobotomized by Dave.

Since when, do you think? I personally think that they're better than ever. On the other hand, I think they moved to more deep learning approaches over the last 3-5 years (based on their research output at least) so perhaps recommendations are perceivably different since then

I’d say more within the last 3. Definitely in the last 2. It’s much worse for me, just repeats the same channels and categories.

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#154

Post by jxramos »

powersnail wrote: post #149
bsder wrote: post #146

Earlier quoted context omitted.

> What's the point of a recommendation system if it doesn't even recommend new video of the same category I like? Because YouTube doesn't care what you want --YouTube cares what it can monetize . I suspect that YouTube gets very little money from showing you classical performances but gets lots from showing you the latest "Funny Cat Video". You are the product , not the customer. Once you internalize that, stuff that…

The thing is, it doesn’t even show me funny cat video anymore. It’s the same classical music video recommended over and over again. To be honest, the more I think about it, the more I don’t know what money is YouTube trying to earn. It’s probably just a dumb algorithm.

I think it has a grasp of revisits, I forget the exact terminology but if you rewatch stuff I tries to predict will you like to watch it again. Some stuff is not worthing multiple times though, once is enough. But music videos can indeed be watched multiple times.

I think I heard somewhere that a lot of kid videos rank in the top stratosphere of view counts on YouTube because kids just play stuff on repeat. There's probably a lot of repeat plays going on across YouTube.

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#155

Post by drivebycomment »

friendlybus wrote: post #112
dexen wrote: post #10

Earlier quoted context omitted.

Systemic fixes are needed. In the meantime you can improve your own experience (and hopefully others a bit too). When you hover over a recommendation, there's a menu icon to the right side (three vertical dots, gray). Clicking it opens menu with option to indicate "Not interested" and "Report". The former is supposed to train your own instance of recommendation algo, the later does the usual reporting. According to t…

Not interested doesn't work. One or two weeks later it 'forgets' your preferences and someone else uploads the same content as a clip so it gets suggested too. Googling around shows there used to be a 'Don't recommend this channel' button which would be awesome to have back. I find it very annoying listening to a two hour debate, understanding difficult topics and then disagreeing with the speaker and being unable to…

> Googling around shows there used to be a 'Don't recommend this channel' button which would be awesome to have back.

It's still there on my Chrome desktop or Youtube app on android.

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#156

Post by SiempreViernes »

heartbeats wrote: post #123
untog wrote: post #105

Earlier quoted context omitted.

> YouTube recommendations can be delightful, but they can also be dangerous. The platform has a history of recommending harmful content — from pandemic conspiracies to political disinformation — to its users, even if they’ve previously viewed harmless content. What part of this is untrue? And again, you're saying "getting rid of it" without evidence. It's a study. It does not mention banning anything, anywhere. If th…

> What part of this is untrue? The characterization of pandemic conspiracies and political disinformation as harmful. It's like kids and allergies. You never hear about kids who grew up on farms being allergic to animals - it's always those whose parents didn't have any when they were growing up. If people aren't subject to misinformation, they'll never develop the sense of who's lying and who isn't. It used to be th…

"I never heard of it happening, so it is universally true that it never happens."

Meanwhile, the actual state of affairs is[0] that there's one allergic farmer child per three allergic non-farmer children. Don't believe everything you read online.

[0]: https://pubmed.ncbi.nlm.nih.gov/11048766/

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#157

Post by retox »

encom wrote: post #106
sigstoat wrote: post #100

Earlier quoted context omitted.

> They could probably infer that if I'm watching comedy ... I don't want to watch some guy's channel about bassists that they've tried to get me to watch 100 times prior. davie504 is basically a musically-themed comedy act, so this doesn't seem that unreasonable.

Urgh him . I watch a lot of music videos (as in videos about making music + music theory). I can not get this guy off my recommends. He has the personality of a sofa cushion. I cannot understand why he's popular. Other than that, I'm pretty happy with my recommends. I think the Google spooks have built a good file on me.

I use this firefox addin, it put a cross next to user comments and videos and lets you hide videos and comments from those users/channels in the recommended videos, searches etc;

https://addons.mozilla.org/en-US/firefox/addon/youtube-clean...

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#158

Post by gothroach »

klmadfejno wrote: post #70
jack_pp wrote: post #38

Earlier quoted context omitted.

They stop appearing if you specifically mark them as not interested

I do this, and it does work, but it doesn't seem to learn that I don't want to see generic youtube garbage. I'd love for it to learn that anything with a shocked face thumbnail and CrAzY question for video name is not desired.

Shocked face or a big red arrow pointing to something in the thumbnail make the video an instant pass for me. It would be amazing if YouTube's algorithm could recognize that.

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#159

Post by 1vuio0pswjnm7 »

powersnail wrote: post #131
FiReaNG3L wrote: post #5

What's impressive to me is that YouTube Recommendation are so ... stale. All it recommends to me is content from the same 3 categories, and the same videos over and over and over (you would think if I didn't click on the suggestion in the past 6 months it would adjust). Now the topics it recommends to me are non-political, but I can very well see that if you fall down one of more extremist rabbit holes thats all you…

Youtube's recommendation algorithm is just dumb, in my opinion. At least half of the time on Youtube, I watch classical music, and you would think Youtube would recommend me some classical performances that I haven't yet seen. Nope. It keeps feeding me exactly the same videos I've already watched, despite being basically the largest catalogue of classical music videos. What's the point of a recommendation system if i…

The recommendation system is not "dumb" in the opinion of its designers. Perhaps it is performing exactly as intended.

In the same way that online advertising needs to display millions of ads to reach only hundreds of people, the recommendation system may not be intended to work at 100% efficiency.

YouTube "recommendations" work as far as I can tell. If you take two people searching for some subject area on YouTube, and only one of them knows how to search past the initial "popular" recommendations, each person will produce very different results.

If neither person is a skilled searcher (most people are not), they both end up with the same results. Those videos receive thus more and more views. Videos with millions of views help Google sell online ad services. If views were distributed more evenly across all videos, would that make advertisers more interested or less interested.

What's the point of a recommendation system that does not recommend a new video in the same category. To promote certain videos to attract large audiences and thus draw more interest from potential advertisers. More advertisers bidding on fewer popular videos would seem to favour higher prices paid to Google.

Some of the best videos I have seen on YouTube have less than 500 views, some even in the single digits. I found them through automated searches, not "recommendations".

Google makes information accessible with the belief that relevance is directly proportional to popularity. The company was founded on this idea.

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#160

Post by gundmc »

duskwuff wrote: post #103
srean wrote: post #90

Earlier quoted context omitted.

How can I get myself on the better side of their A/B test ?

Unfortunately, I don't think there's any (public?) way to force it. I've seen the feature disappear or reappear on a reload -- it's not consistent across a session, let alone on a user-by-user basis. Try refreshing the home page until it shows up, maybe?

At youtube.com/new, they say you can opt-in to experiments if you have premium, but I don't have premium and am not sure if that is one of the options.
Post reply on HN