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

YouTube’s ‘dislike’ barely works according to new study on recommendations

engadget.com

Post reply on HN
Print view View original

321–330 of 437 posts

Re: YouTube’s ‘dislike’ barely works according to new study on recommendations

#321

Post by robertlagrant »

lemon_bottle wrote: post #238
TheAceOfHearts wrote: post #32

The trick to improving YouTube recommendations is to regularly prune uninteresting videos and channels. Eventually this should start surfacing videos you're more likely to find interesting. I've actually hit the end of my recommendation stream on the mobile app after marking dozens of videos with Not Interested. Once this happens it starts recommending videos you've already watched or older videos from creators you'v…

>> The trick to improving YouTube recommendations is to regularly prune uninteresting videos and channels. Eventually this should start surfacing videos you're more likely to find interesting. Unfortunately, Capitalism doesn't work like that :-(. Back in the old days, Google searches used to be organic searches in the genuine sense, and that was the best Google we ever had. But today's Google is the Big Tech Capitali…

Google was always capitalist. No one was forced to work on it or invest in it.

On your other point: I've not read anything that suggests Google is paid things to promote content. Where do you get that from?

Re: YouTube’s ‘dislike’ barely works according to new study on recommendations

#322

Post by dredmorbius »

eimrine wrote: post #4
PaulHoule wrote: post #3

It’s been a while since I’ve logged into YouTube but my impression is that the experience including recommendations is worse if you are logged in.

Best YT experience is to clear cookies then open a lot of YT links from HN, but when our human nature drives us watching a stupid video or two from YT recommendation then after that session do clearing cookies and open YT links from HN again.

Using YT via Incognito or Firefox containers, with frequently-cleared cookies, and using search as a primary discovery method ... works OK... It's not great, but it's not terrible.

The recency bias means that whatever terms you're searching become prevalent for that session. It's not possible to dismiss items or channels (idiotic on YT's part IMO), but you can at least create a short-lived preference instance unpolluted by any previous explorations.

Re: YouTube’s ‘dislike’ barely works according to new study on recommendations

#325

Post by dredmorbius »

jiggawatts wrote: post #117

What a lot of people are missing is that when watching YouTube, they're the product, not the consumer. I know this has been said many times before, but ignoring this reality is like trying to understand life without considering evolution, or understanding chemistry while ignoring the periodic table. Google's incentives are to tailor customers to advertisers , not customers to content! They don't care if you see what…

Annoying a sufficient fraction (or influential cohort) of the product off the platform will eventually have negative impacts.

Google crossed that Rubicon 10--15 years ago. They're driving with increased speed into the other side.

Re: YouTube’s ‘dislike’ barely works according to new study on recommendations

#326

Post by Garvi »

About 6 months ago I ran my own little unscientific experiment with Youtube, after I noticed them promoting a certain channel that I've never watched. Nothing political or anything like that. Just a run of the mill youtuber, currently very popular with the kids.

The first week or two I was pushing the "not interested" and "don't recommend channel" every time one of the videos would show up in my feed (mind you I've never actually clicked on one and watched it and I'm a firefox+ubo user). Those two buttons had no effect at all. So I tried flagging the videos for "unwanted content". No effect either. The videos would keep on appearing in my suggestions at various places for a whole month straight.

The theory being that the channel was being actively promoted by Youtube and was thus impossible to hide. Ties in with the removal of the 'dislike' button as I believe it had the same reasoning behind it: raising the value of their advertising service.

Re: YouTube’s ‘dislike’ barely works according to new study on recommendations

#328

Post by mattbrewsbytes »

beart wrote: post #6
HardwareLust wrote: post #5

YouTube doesn't seem to understand my preferences, at all. It's far, far too biased toward what you recently watched. It basically doesn't work and certainly doesn't "learn" anything.

I feel like every recommendation service works this way - YouTube, Amazon, Spotify, etc. Whatever the last thing you looked at becomes your entire life's mission according to these systems. When I think about all the buzz words around machine learning, ai, and so on, it's completely incompatible with the reality I see. What exactly are all these engineering teams doing just to end up with this result? Amazon might be…

It appears to any end user that knows some basic logic and looping concepts that the advanced AI/ML/whatever recommendation engine is no better than looping over the list of recent content (videos, products, etc.) and selecting more content that has same keywords. If you remove the volume and distributed systems elements for large systems its probably a single SQL query joining 2 tables, history and products (or videos or whatever).

For the bike rack example, to make it "smarter" you could have a product type that is something like "one-time purchase" and have a linked product type that is accessories for the one-time product, like helmets, pumps, tire changing kits, inner tubes, water bottles, etc. This would be what a retail worker at a bike shop would do - ask if you need any accessories, maybe point out clearance/sale items. Thats it, no fancy engines/AI/ML needed. If the customer wants other stuff, there is a search bar they will likely use.

I chalk it up to Resume Driven Development and just the innate tendency of engineers over-engineering things because that's how our brains work.

Re: YouTube’s ‘dislike’ barely works according to new study on recommendations

#329

Post by thelittleone »

Recently I saw a video on my youtube feed claiming to be a "Live" crypto discussion with 4 people including "Elon Musk". It looked pretty legitimate so I checked it out. It was obviously pre-recorded and re-streamed and a scam. So I flagged the video and 4-5 hrs later got a notification that it has been removed. The next day the same video appeared. I flagged it again and it was removed 4-5hrs later (note there was 50k+ viewing the live stream). The next day it appeared again. I stopped bothering to report it.

Does youtube not have a responsibility to protect viewers from fraud? Surely they can detect the same video being posted again?

I wonder how many people are losing money to these scams.

Re: YouTube’s ‘dislike’ barely works according to new study on recommendations

#330

Post by mattbrewsbytes »

amatecha wrote: post #20
HardwareLust wrote: post #5

YouTube doesn't seem to understand my preferences, at all. It's far, far too biased toward what you recently watched. It basically doesn't work and certainly doesn't "learn" anything.

Yeah, on a non-logged-in system I watch YT on, I watched exactly ONE gaming-related video, and now suddenly the "home" view is full of Minecraft, Roblox, whatever FPS shooters, etc... Like... I'm not suddenly absolutely obsessed with a topic because I viewed one video, guys...

It operates similar to saying to a spouse or in-laws or other family that you like some and they are the type that runs with it until you have to awkwardly tell them "No more". Every holiday, birthday, random "oh I saw this and thought of you" type of thing.
Post reply on HN