>TikTok only needs one important piece of information to figure out what you want: the amount of time you linger over a piece of content
Why have YouTube Shorts and Instagram Reels failed to implement this? It is the main reason the TikTok algorithm learns so fast and is so much better that their algorithms. The other important reason is that the "Not interested" action is seamless and TikTok actually learns from it.
The problem that Youtube has is that they already were what TikTok is today. Early ~2005 Youtube was extremely similar to what modern TikTok is, lots of amateurs talking into their webcam and not a whole lot of professional content.
Youtube however spend a lot of effort to drive that early content away and focus on professional content that they could monetize easier. Finding a random video from somebody who isn't a professional full-time Youtuber is getting pretty rare these days.
Youtube Shorts thus has the problem that all the random slice-of-life content that strives on TikTok, is deliberately suppressed on their own platform. Difficult to see how they can fix that, as it just goes against everything they have been doing for the last decade.
The algorithm is important of course, but it needs content to search through and I just don't really think there is enough left on Youtube to compete with what TikTok is doing. All the videos I saw on Youtube Shorts was just professional Youtube's cutting some bits out of their larger videos into a shorter video, it just felt useless, annoying and uninteresting.
> Youtube however spend a lot of effort to drive that early content away and focus on professional content that they could monetize easier.
Could you elaborate a bit on that? What did youtube to do drive that early content away? Why is it easier to monetize professional content?
I have an alternative theory: Some people within every social media company try to turn the company into a mass media company—they want to promote content according to their taste, hang out with celebrities ... See "youtube originals". Social media companies win if they can withstand that pressure, if they manage to stay neutral, to show things to people that they're interested in.
> Why is it easier to monetize professional content?
For the same reason that Twitter is currently under fire: brand safety.
Brands don't want their ads to appear next to people spreading flat earth conspiracy theories, drunken ramblings, hate speech or dangerous "challenges" - just think back to the Tide Pod "challenge", if you were an advertising buyer for Tide, would you want an ad for Tide as a pre-roll for a video where a bunch of morons eats them like candy? That crap alone led to almost 7700 hospital incidents and at least fourteen deaths [1]. Additionally, you have people without a fucking clue what they are doing who upload videos like high voltage arc burning in wood, showing no concern to safety - and at least 30 people died because they were blindly following these videos [2]. Other stuff I've seen are "DIY" YouTubers doing electrical work and even I as someone who's not formally certified but have way over a decade worth of experience in construction and maintenance immediately see just in which amount of danger these people go.
And hell even actual professional YouTubers like HeavyDSparks routinely show a shocking lack of caring about workplace safety. Not wearing proper PPE (helmets, shoes), standing barely a foot away from a winch cable with thousands of pounds of load on it... it's bad enough that people don't care about OSHA guidelines on their own, but showing that on YouTube... I respect the guy and his team for the good work they do to help out people and nature, but they could definitely do way better in promoting safe workplace standards!
Platforms that allow large scale unmoderated publication of user-generated content will always end up facilitating grave injury and death.
> The other important reason is that the "Not interested" action is seamless and TikTok actually learns from it.
Have you ever used it? Because is doing a "decent" job but not a "great" job. I've been suggested videos that I constantly flag as "not interested" all the time (and since it has no "block" functionality you gotta suck it up if it comes back again)
The algorithm is great but is more about where are you located than anything else. You don't get to see a lot of the content created in the first place just because you're not in the right region (eg. I never saw a Charlie video for example)
Overall I think it might apply a weighted approach on what to show and based on your engagement double down on what you are likely to watch more. Then you've trends, top creators etc that get featured more often.
The only thing I love about TikTok is the non-existing advertising. There's some, but is rather seamless and non blocking that sometimes you are even going to watch it.
Nothing to do with YouTube where you're basically in a worst situation than the classic TV (ads every now and then)
My very loose understanding is that the effectiveness of the "Not interested" button is somehow inversely proportional to the strength of the correlation that TikTok uses for that particular suggestion.
As the obvious example, if you are a male, you might only need to click "not interested" once or twice to stop seeing e.g. skateboarding or woodworking content, but many, many times to stop seeing content from attractive females.
I suspect it 'knows better than you'. Self-curation leads to a boring niche whereas ignoring what you want in favour of what you linger on - even if you're lingering out of rage - leads to more engagement.
I'm not convinced the Not Interested button does anything at all. I have blocked and "not interested" 103 (one hundred and three) reddit accounts on tiktok so far and it still shows me them.
The actual interesting information is one meta level above.
User attention deformability.
How succeptable is the user to suggested content, that allows for faster "steering" into addiction lanes.
So, what it basically is, a search over addiction topology to the first hook and then a try over sorted by addictiveness topics for similar classfied user groups.
TikTok has paid human curators iirc, I don't think the others do. This lets them be more aggressive with recommendations since the videos it recommend are more accurately labelled.
Humans doing curation is not that expensive, that is the main innovation TikTok came with.
Exactly this; TikTok has careful moderation and curation within specific styles and a (relative simple) algorithm to quickly figure what the user is interested in. Plus a lot less advertisement.
That is their "innovation"; and is also something that is hard for YouTube and Instagram to copy as it would mean they would have to spend more and monetize less.
A popular theory I’ve heard is that it’s quite common for social media platforms to “turn up distribution” in the early years to attract content creators, but that they’ll inevitably turn it down later once they think creators are significantly reliant on the platform. I guess the idea is that this short of rapid distribution of content across the network and learning what consumers like is potentially at odds with maximizing revenue?
> >TikTok only needs one important piece of information to figure out what you want: the amount of time you linger over a piece of content
I thought this is the naive view expressed to non-technical people, but it incorporates a lot more user input. It's just, to the user, it feels like that's the only thing it's communicating.
Sure, but getting this "lingering" metric from 100% of its users already puts them leaps ahead of other social networks. Every user provides clear interaction metrics for every content he sees.
On the "classic" consumption UI's of other social networks you couldn't create that metric with such high accuracy, because you don't know which content is in focus and the content itself might be static.
That's probably also why Instagram's algorithms now put much higher weight on Reels than their classic feed when measuring engagement of a user...
Perhaps it's not as universally worthwhile as it seems. It requires length of attention and preference to coincide, and though anecdotally, I know it doesn't for some people - like myself.
I tried TikTok in my usual manner and attitude towards social media; I would doom scroll as fast as I could, allotting longer time for patterns I don't recognize, shorter for what I already understood, tangential to preference and couple seconds at most. The feed or whatever it's called became monotonous, lacking in diversity, devoid of content, even those that I would be presented with in other platforms as trending on TikTok. And I deleted the app.
YouTube isn't as terrible, though not ideal(perhaps my overuse of ad blocking is playing a role with it).
I watch about 10-15 hours of TikTok a week and agree that the algorithm is incredible.
But personally, I haven’t found TikTok’s “not interested” button to be useful at all. It’s like it puts content/creators in some limited time out and then brings them back with a vengeance.
On TikTok I’ve found the best thing to do is block creators who I’m not interested in. It’s the only surefire way to keep them out of my feed and is very effective.
Why have YouTube Shorts and Instagram Reels failed to implement this? It is the main reason the TikTok algorithm learns so fast and is so much better that their algorithms. The other important reason is that the "Not interested" action is seamless and TikTok actually learns from it.