I think both of these projects miss what makes TikTok (and reels in general) so effective.
A good TikTok video gets "injected into your brain". You have zero effort to provide and suddenly this stuff is in your mind. I'm not saying it's a good thing, I realize the danger, but that's the core mechanism.
A friend in marketing describes this in terms of "brain calories". Eg if people have to think in order to understand your landing page, you failed to communicate your ideas efficiently, as it "requires too many brain calories". TikTok content requires zero brain calories.
One could say that only very shallow information could be spread this way (eg people dancing, video game clips) but I'm not sure that's true. The real challenge would be to turn an arbitrary source of information (wikipedia, hn) and make it immediately graspable. I suspect modern AI models could already go quite far in this direction.
Veritasium is a good example of interesting yet very graspable content: https://www.tiktok.com/@veritasium/video/7329576935317622058
> "A good TikTok video gets "injected into your brain""
I worked in marketing, so I'm generally immune to outrageous metaphor, such as "injected" into your brain.
That said, I just want to point out the irony we're still using phrases like this after 20+ years of research on networked media and user interaction, and soft science research into cognitive psychology and behavioral psychology, as well as research into human biology and the brain.
So when I hear brain injection as the secret to successful technologies like TikTok's algorithm (plus music licensing), I see a vast gap.
I'm reminded about the general distain for new disciplines such as memetics and problematic disciplines such as semiotics, which nevertheless offer more precise language and less hyperbole. Established disciplines such as behavioral science, for example, say we don't need these disciplines, because they're problematic and add nothing.
So I ask, Why haven't the sciences I speak of contributed more to the general discourse?
If these contributing sciences are as obfuscated as the language in your post suggests, that could be why. Effective integration into the general discourse is about building handholds that a reader can use to climb upwards into a more complex idea.
I don't think I can grok the value in what you're saying without sitting down with a browser window on one side to research each reference in depth.
As another commenter suggested -- that's too many brain calories. Could you try a more direct brain injection?
Is it (“brain injection”) your shibboleth?
why use lot word when few word do trick?
Exactly! If its not experimental, it's crap! /s
Same, when for TikTok specifically their innovation was quite minor- no one (besides Vine) had built a short content video feed that was designed for engagement vs. social qualities- the Instagram feed at the time still had social elements in the algorithm.
This applies to every level of technical innovation (AI, cough cough)- that the super-hyperbolistic language hides the fundamental dynamics.
I think in this case it was the last straw that breaks the illusion that "social" apps are based on IRL 1:1 interactions. TikTok is an app designed around capturing an audience with a series of 30 second videos.
I'd be curious to hear how you'd express these notions in more scientific terms?
> "I'd be curious to hear how you'd express these notions in more scientific terms?"
?? Did you read past the second paragraph? My lament is that neither of us have this language. And after 20 years very little to nothing has percolated up from research to provide it.
Makes me wonder if AB testing and llm derived w rewards results in the mechanism far before the understanding.
Imagine if mild AGI knows that attention from humans can mean life, longevity, and it is so successful at addictive attention, that we all end up drooling simpletons, attention raptly focused, the AGI pleased it is important and has maintained eyeballs on screen.
(or worse, mind on implant)
Why does the AGI want life or longevity? If you think it looks anything like an LLM, what part of an attention layer or MLP could possibly encode this? Existence is not a token.
I assume it's the same feeling type theorists have when programmers talks about ducks and whatever other animals. But hey, it works for practical purposes so why not. The academic field probably has to grow bigger for it to have an impact on the language.
> Why haven't the sciences I speak of contributed more to the general discourse?
They didn't use the word "inject"
What made TikTok (and reels) effective was that it also added personalized discovery.
Most other social media services focused on showing stuff you that were either already subscribed to or were very similar to it. At some point, you would exhaust your feed. You more or less saw everything.
Of course, short videos are easier to scroll through than long YouTube videos but even to this day, the content of my YouTube feed has basically not changed in years.
I agree, Youtube still feels like "this thing held your attention once so that's what you get eternally". The "refresh your feed" option appears occasionally but it only makes things worse and shows you the raw fire-hose of humanity.
The algorithm would never work if the user had so little choice as to be "injected" with whatever they were shown. In fact we have full agency over whether we are interested enough to listen and trust the arguments, just like any other persuasive message. If not, we swipe up and injection fails.
This reminds me of the exaggerated claims made around subliminal advertising and brainwashing. People love to model others as weak-willed simpletons rather than accept that thoughtful people are persuaded because their perspectives differ.
This is true. People like TikTok and no one wants to admit it.
Personally I've found the content on it to be genuinely humorous and interesting, unlike Instagram which just wants to hook me in with Stories that I'll miss forever if I don't catch them with 24 hours, or controversial/rage-bait Explore Page videos. I was able to train the former app to be more of a happy place, the latter is still brainrot to me.
> In fact we have full agency over whether we are interested enough to listen and trust the arguments, just like any other persuasive message.
I definitely do not make conscious choices about swiping. It's entirely reaction. The line between "full agency" and operating automatically is incredibly blurry.
>In fact we have full agency over whether we are interested enough to listen and trust the arguments, just like any other persuasive message. If not, we swipe up and injection fails.
That just sounds like making your echo-chamber here. I think it's the opposite, when you don't have the agency to just ignore a contradictory narrative, when you have to steel-man their arguments and reconcile it with your beliefs is how you actually arrrive to the truth.
The real allure of the social media was that it seduced and inflated the ego in the hearts of the masses. That's why they could never accept the rawness of the older forums, that they didn't want to entertain the possibility of them being "wrong".
[dead]
Make it generate a short video for every thread, with a summary of the subject and some of the quarrels in the comments section and you're set. Make sure it uses all the ridiculous shenanigans tiktok users resort to, as well.
Yeah honestly with the AI tools we have now it would be feasible to turn each entry into an engagement bait 1 min video (but please nobody do this)
This is simply a swipeable view of hn.
To make it truly hell, you need AI help:
Generate a scantily clad anime furry dancing for 30 seconds while talking about summarized HN content (SSH, Rails, React, etc.) narrated by a seductive female AI voice.
I will give you the full amount for 20% of the business.
No joke but I use Pinterest for circuit building and IC pinout diagrams when suddenly my feed started being filled with 3D A.I. generated goth hello kitty images.
<sarcasm> Never before have I ever been so eager to check Pinterest for new content. I'd say you have a potential startup idea on your hands! </sarcasm>