This isn't possible on Youtube right now. The automatic tools for detecting LLM-generated content have far too many false positives. And obviously no one is going to pay an army of people to curate the content. The best thing right now is to rely on the reputation of individual channels that you are personally familiar with.
Youtube's automatically applied label will be worse than useless unless they've made some remarkable breakthrough, which I doubt. They'd be better off just using creator-applied labels, and of course if they would label anything that Youtube itself contaminated with automatic translations or its ilk, that would be good too.
Maybe they could hand out lifetime bans to people who upload untagged AI music? Obviously that wouldn't eliminate the problem, but I could see it helping.
Well, theoretically you could build a service providing blocklists, and users could subscribe to such blocklists with a browser extension blocking accounts. Basically Sponsorblock or Blocktogether for Twitter, with individual users flagging accounts for slopaganda, content theft, rage / engagement bait and other issues.
Unfortunately, it's way, way too likely that you'll run into some sort of bot detection on Youtube's side and I've seen more than enough horror stories about people getting fucked over and getting their entire Google account perma-banned with no way of recovery.
It's clear that YouTube doesn't want you to have much influence over your feed. You can't even ban specific channels from being shown to you, which would be the simplest thing to implement, and other knobs that previously existed were silently removed.
Since Google does nothing that isn't based on metrics, we can deduce that they have data to show that giving people settings to focus the recommendations on what they want reduces total watch time. We'll only get an AI filter if it turns out that AI slop offends people so much that they disengage with YouTube altogether, which outside of HN and similar bubbles, I don't yet see happening.
> You can't even ban specific channels from being shown to you
Yes, you can. Click the video's 3-dot menu > Don't recommend channel. Though I have noticed that this only blocks them from showing up in the feed, not in the recommendations sidebar. I also have to run uBlock to hide shorts, already-watched videos, subscriber-only stuff...ain't saying the YT experience is good, not by any stretch of the imagination.
You can click a button that makes a strong "suggestion" to the algorithim, which they will honor for as long as they feel like.
I went through this a few years ago when the channel of a large far right "news" broadcaster kept being jammed on my front page, and the best I could do was keep hitting the button and have it it "temporarily" be removed from my front page before it would inevitably show up again months down the line.
Perhaps it is not deliberate, and merely incompetence. Either way resolved on desktop with an addon because if I wanted to gamble, i'd go to a casino.
Yes, this is true. I did notice that a bunch of channels I knew I had blocked started showing up again, but in my case it took 2-3 years. If it only lasts months for you, it's much less useful.
To be fair, it is entirely possible it works better today, than it did then. I was just so aggrivated at the time, thinking each time I had resolved it, only for it to appear again that I just gave up!
Yeah, just let me hide all the AI content. Far too often I stumble onto something that looks interesting, and halfway through I realise it's not really saying anything. It's just AI drivel designed to capture my attention and hold it for a while.
When I used to use YT, i used https://untrap.app/, it was a great improvement.
If you think you can't quit youtube, I used to think that way, and then i did it by deleting my account and using libredirect for invidious. my usage went to just a few minutes or 0 per day.
I'm not particularly religious but I did give up Twitter for lent as a test of my self control.
I highly recommend everyone occasionally do this with social media as it was somewhat eyeopening how much better I felt overall. This was mostly due to not being exposed to the doom scrolling you can eventually get pulled into (despite efforts not to).
I did miss feeling like I was "plugged in" to the stream of news/memes etc though.
I've already got more interesting, informative books on my to-read list than I'm going to have time to finish in my lifetime, so I think I'll be fine without youtube.
This is exactly how I consume YouTube as well. I do keep the side recommendations on since they mostly contain music or videos I've already watched, which I don't mind.
I'm now experimenting with hiding thumbnails too, and honestly I've been liking it a lot. It's a very curious feeling how my eyes can no longer latch on to something visually appealing, and instead try to look for information in channel names.
It's somewhat deceiving practice IMO although it could simply be my insecurity.
Along with the empty page, it says "Your watch history is off" in bold then says "... change your setting ... to get the latest video tailored to you"
It sounds as if I'm missing out on latest videos which, technically true, but I wonder if that wording is necessary. It could've just said "Update the settings here to get recommendations". But of course for-profit companies need to make profit :)
Yeah exactly, they could have made their service useful by showing your subscriptions instead. Yet, they decided to enshittify for people who want choices.
Likewise. The page is youtube.com and then just /feed/ without anything else there. That's the blank page, thank goodness they've not ruined that yet :)
Doubt it is best. Taiwan China and Singapore easily better than SKorea. Singapore is more unique where everything is resources tight they still able to create that system.
> Have you observed Bun have more segfaults, OOMs, etc, since the Rust rewrite? Have you noticed more security vulnerabilities? Have you seen more bugs? (Of course you haven't, the rewrite hasn't even landed yet.)
Your argument could go other way too. Why haven't they landed if they're so confident with the change?
The idea is similar to maintaining on-prem vs cloud
Cloud is optimized for development velocity but its nature of high margin business eventually makes on-prem more promising
It could be too late but it might be worth looking into tax saving if you have a business. Depreciation of asset is a loss and may deduct your income. (I'm NOT a tax expert)
Cloud servers have cheaper electricity, the scale of industrial-level cooling, no issues for you (as a user) with hardware failure (ie you just use a different server; it's not your problem) and can amortize their cost by running 24x7. I've seen H100 computer hours for as little as $2.
As the author notes, there are also electrical/wiring issues that cap how much compute gear you can run in a space not designed for it. I suspect a standard 20A 110V circuit can probably handle 2x RTX 6000 Pros. 15A probably can but that requires more research. Anything more than that and you're using multiple circuits, which has issues, or you need an upgraded circuit (eg 40A 240V) with all that entails (eg heavier duty cables, custom plug, etc).
I suspect a standard 20A 110V circuit can probably handle 2x RTX 6000 Pros. 15A probably can but that requires more research.
During initial setup of the server I am putting together, I found that a machine with 4x Blackwell cards derated to 300W can get by on a single 120V 20A circuit. It's tight but doable. A lot depends on the power supply. I don't think it's a great idea to run 4 high-power GPUs on a single ATX-style PSU, even a beefy 1600W job.
The other questionable part is whether all four cards can temporarily spike at full power during boot, before the wattage limit is applied by the OS. Some accounts say this is possible, and if so it could shut down the party in a hurry. But I didn't see any misbehavior when I tried it.
My earlier research suggests NVIDIA does not actually cap spikes, it caps the average over short periods of time. So setting the power limit is no guarantee.
Perhaps Github, by default, should add its own .gitignore that ignores files with certain keyword and have it only allowed to override by repo setting.
I've seen too many incidents when an engineer checks in a plaintext password to a repo
reply