randomblock1

There’s a button “Show Bookmarklet and Settings Data” that saves all the settings to query parameters

Oops. I meant the modern method is better if network is reliable. Not unreliable

This seems like a silly workaround at first but it’s really not. If the network is unreliable, you can’t really use normal video streaming, you need to send full screenshots. Probably still a better idea to use only I-frames than a bunch of JPEGs but whatever.

But they did make some very silly mistakes. Par for the course of an AI coding company I guess.

1. WTF are you doing with 40mbps. Tone it down to like 8.
2. If the network is reliable but slow, just reduce bitrate and resolution. Don’t use JPEGs unless packet loss is the problem.
3. WTF are you doing using a whole game streaming server for? It’s meant for LAN, with minimal latency. Just capture the screen and encode it, send via WebSockets. Moonlight is completely unnecessary.
4. Only keep the latest frames on the server. Don’t try to send them all immediately or it’ll fall behind. Wait for the client to finish receiving before sending another one. This way it won’t lag behind increasingly. This should have been extremely obvious.
5. H264 is so 2003, ask the client if it supports AV1 or HEVC then use that, more data for free.
6. Use WebTransport when available, it’s basically made for live streaming
7. Why are you running a screenshot tool in terminal then grabbing the jpg… Unnecessary file overhead & dependency

I probably missed some but even for an AI company this is really bad

The modern method is more efficient if your network is reliable.

Modern: I-frame (screenshot) then a bunch of P-frames (send only what changed). If a frame gets lost, the following frames aren’t really usable, frozen until next I-frame Old: Send only full screenshots. Each frame has all the data, so losing one doesn’t matter.

Why ipv6 only though? Is there something about it that makes it more resilient to DDOS? If a device on the botnet has both ipv4 and ipv6 I don’t see how it’s mitigated

Just install fd-find. I found it to be the most intuitive way to search for files

https://www.researchgate.net/publication/304572591_A_Meta-Analysis_of_High_Resolution_Audio_Perceptual_Evaluation

“Results showed a small but statistically significant ability of test subjects to discriminate high resolution content, and this effect increased dramatically when test subjects received extensive training.”

Basically, people can just barely detect high res audio but it’s not much better than a coin flip. If you have lots of experience you’re more accurate but not by a whole lot.

Anyway 48kHz sampling can produce up to 24kHz and the human limit is like 20kHz. Most songs don’t have 96db of dynamic range, and 120db is hearing damage, so the idea that the average person can easily hear the difference is not true.

You could make that case for the boomers, but with the advent of the Internet it’s not really limited to specific countries anymore. Gen z just describes people who grew up with some access to the internet, but not to the extent of alpha. A large part of these recent revolts were caused or accelerated by social media. Gen Z is more specific than young people, and in this case also more culturally relevant.

Doesn’t matter anyway, even if you think Gen Z isn’t real you know what age range the article is talking about.

This was desperately needed.

But I’m not convinced they aren’t just going to make alts.

Isn’t there already a special low-power part of phone chips designed to listen for wake words?

I’ve always wondered why there isn’t VR optimized compression. Seems like each eye would mostly have similar or shifted pixels than the other. Nice to see some work being done in this space

Reverse image search. Some guy posted this on Reddit a year ago

Found this, looks pretty close

No, it’s expensive to comply (at a massive scale), but easy to avoid. Just change the user agent. There’s even a dedicated extension for bypassing Anubis.

Even then AI servers have plenty of compute, it realistically doesn’t cost much. Maybe like a thousandth of a cent per solve? They’re spending billions on GPU power, they don’t care.

I’ve been saying this since day 1 of Anubis but nobody wants to hear it.

And when GPT-5 stops doing this, all of a sudden it’s a bug and they want 4o back.

up to 1000x energy consumption efficiency in these workloads

Seems like a win to me

The FDA would probably also like to have a word.

A good example were those Apple AI ads. So cringe. Google’s ads aren’t much better but at least Gemini works.

“How do you read a file?” “Cat, more or less.” ba dum tss

You’re on a forum right now, no? I think a lot of Discords have accompanying Reddits and vice versa, hopefully they start using Lemmy. But forums are still popular. Just not Ye Olde BBS