Live data from Hacker News

Mozilla Is Crowdsourcing Research into YouTube Recommendations

foundation.mozilla.org

161–170 of 179 posts

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#161

Earlier quoted context omitted.

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,…

This isn’t how digital ad inventory works. Advertisers don’t bid to win a slot that all viewers of a particular video will see. They bid to win a view. A million views on one video, or the same million impressions split across a thousand videos doesn’t change the supply/demand equation like you suggest. Buying an ad on the most popular video isn’t like buying an ad on the Super Bowl, because not all viewers are seeing the same ads.

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#162

Earlier quoted context omitted.

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,…

A video with only a few hundred views is not going to produce much YouTube ad revenue. In order to sell the idea of making money from YouTube ads, one needs to convince ad buyers and video uploaders that videos can receive millions of views.

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#163
post #150
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…

> Because YouTube doesn't care what you want--YouTube cares what it can monetize. > You are the product, not the customer. > Once you internalize that, stuff that Google does makes sense. They published their algorithm. They rank items according to predicted watch-time. The other poster is probably correct, in that many people use these long classical music videos as background music over and over (i.e. high watch ti…

Interesting. When and where did "they publish[] their algorithm?"

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#164
post #98

Earlier quoted context omitted.

Except for the memory leaks.

Chrome eats a lot more memory for me than Firefox.

The difference is, chrome allocates what it needs upfront, and that's all it uses. Firefox keeps increasing its usage.

That means that when I want to leave a website open on my server for a month, I use chrome because I know exactly how much memory it's going to use. If it was firefox, at the end of the month even the swap would be filled.

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#165
post #38

Earlier quoted context omitted.

I use YouTube infrequently. Usually for binging random content like comedy sketches or speedrunning. No matter what I'm on, there's always a few videos in my recommendations that make no sense and look like trash. Not generically, but specific videos that ALWAYS appear. One of them is "I PAID FIVE BASSISTS ON FIVR TO PLAY AN IMPOSSIBLE BASSLINE" ... Why is this so bad? What are your hundreds of data scientists doing?

They stop appearing if you specifically mark them as not interested

I've marked dozens of videos on a single topic as "not interested", only for them to be replaced by different videos on the same topic soon after. Marked all those as "not interested" again, and the next day there were more.

The only thing that works for me is finding the video in my watch history that's causing the recommendations and removing it there. Often it's even something I didn't watch all the way through or even gave a thumbs-down; apparently thumbs-down also means I secretly want more of this content?

YouTube's recommendation algorithm is somehow worse than useless.

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#166
post #151
post #26

Earlier quoted context omitted.

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...

> you can select a category

Could you tell me how ? Would love to have that feature. They seem to have turned this off.

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#167
post #166
post #151

Earlier quoted context omitted.

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...

> you can select a category Could you tell me how ? Would love to have that feature. They seem to have turned this off.

I'm sorry I don't know how to turn it on. When I go to youtube.com, at the top it has about 15 categories that it has gleaned from my use (e.g. Guitar, Flight Simulators), some of which I'm surprised it learned, as well as All, Mix, Trending. I use Edge and I'm logged in to YouTube.

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#168
post #150

Earlier quoted context omitted.

> Because YouTube doesn't care what you want--YouTube cares what it can monetize. > You are the product, not the customer. > Once you internalize that, stuff that Google does makes sense. They published their algorithm. They rank items according to predicted watch-time. The other poster is probably correct, in that many people use these long classical music videos as background music over and over (i.e. high watch ti…

Interesting. When and where did "they publish[] their algorithm?"

RecSys 2016

Covington, Paul, Jay Adams, and Emre Sargin. "Deep neural networks for youtube recommendations." Proceedings of the 10th ACM conference on recommender systems. 2016.

Nvidia furthermore implemented/released it for TensorFlow, and for their new Recsys engine:

https://ngc.nvidia.com/catalog/resources/nvidia:wideanddeep_...

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#169
post #147

Earlier quoted context omitted.

I think the "exclusively for Firefox development" is pretty important. Mozilla is doing a lot of random things these days

>Mozilla is doing a lot of random things these days Yeah, like paying three million to its CEO annually. That's not going into Firefox development alright.

The one who no longer works there? And how do you think they would build and manage a development team without executives?

Re: Mozilla Is Crowdsourcing Research into YouTube Recommendations

#170
post #164

Earlier quoted context omitted.

Chrome eats a lot more memory for me than Firefox.

The difference is, chrome allocates what it needs upfront, and that's all it uses. Firefox keeps increasing its usage. That means that when I want to leave a website open on my server for a month, I use chrome because I know exactly how much memory it's going to use. If it was firefox, at the end of the month even the swap would be filled.

If you try to open something else after that month, does your computer crash or does Firefox happily give up the memory?
Post reply on HN