booksfilmtelevisionmoviesmusicmusicvideosgamingoldvideosfoodcookingpoetryartphotographytravel80s90s2000s70s60s50smetalclassicalanimationfilmmakingcartoonsmusiccoversTop Subs
4

I like most people have a bit of a youtube addiction. But youtube is starting to make it easy to quit.

It was once the case that most videos on youtube were around 4 minutes long. Long enough for someone to share something genuinely unique, but quickly, and without putting strain on them to edit something really significant. The ideal length was anywhere from 4-15 minutes.

But then youtube tried to kill videos in that range to encourage longer watch times per video. So youtube became long form media analysis, which sometimes would come close to overshooting the media in question in length. Often this would be someone who didn't know that much because they are forced to produce this high volume on repeat, and edit it, and so they wouldn't have enough research to fill that time. So this content became padding and talking in loops, sharing their core opinion across all videos, stating it 40 times per video. It was bad.

Youtube realized people don't all want to watch that. Not everyone can commit an hour and a half to watching a video on why Colgate toothbrushes have some bristles colored. But they can boost total screen time instead of individual per video screen time with a scroll down format. But it was too short. There was no time for context in 5 seconds. So they kept expanding the time.

So now it's mostly multiple clips pushed together with stock reused reaction cuts that don't even match the content, AI, and even reactions to AI, and AI reactions to not fully coherent cuts of real content. Basically the laziest thing someone could make a script to bulk create 50 such videos in a single command line. Anything that moves and makes sound is good enough.

Should I do something with gvid, soon becoming uvid, to separate it from this noise? It started as the neutral platform. But something has to be said about differentiating yourself. And I think people want somewhere they can go with real humans. An argument against is uvid doesn't have a good monitization strategy like youtube has. Which means if effort isn't worthwhile on youtube, why would it be on uvid. Counter argument. Maybe it's the monitization on youtube driving over-optimization of bad content. In reinforcement learning we call this exploitation vs fit. Counter counter argument. uvid is likely to not get that much original content, so if slop is made anywhere there is no reason it won't end up on uvid as well. Counter counter counter argument. uvid could start to filter this and be more curated. Not that curated. Just verified human.

Options for how we could do this or encourage this if we decide we want to.

  • Add sort boost to videos 4-15 minutes long. That seems to be a good proxy metric for videos without any of the key attributes of slop. Because anyone who is exploiting rather than pursing authentic fit would be targeting a different time. That's not to say good content can't exist out of that range. But it would function a bit like fashion where those whose interest runs counter to a bad idea know that taste oriented people are using that as a proxy metric.
  • Add a check box at the top of the suggestion to sort boost that range on the client side. Literally the same suggestions on the page, but the ones in range at the top.
  • Add a toggle to flag content as AI dominant, and a check box to ask to hide that content. I would want this to be outside of the tag/keyword system of uvid because content about AI would otherwise get mixed up.
  • Most aggressive would be manual/automatic removal of AI flagged content over a week old. Let's face it, most of that content doesn't need permanent vestige. Plus, for that content it's likely a youtube embed, in which case it is preserved, for the case posterity really needs the slop.

Since gvid is being re-branded to uvid, should I also re-brand it as a refuge for humans and human content? I could.

Comment preview