/tech/ - Technology

>>34496
> i think it's mildly good news
i said this on the usapol thread, but the burgerreich AI industry is 100% irredeemably fucked, like there's no scenario where they succeed, because OpenAI and Anthropic spend billions of dollars advancing their models an inch on their benchmarks, and every other lab in the world gets to reap the benefits of this massive investment by simply distilling the frontier models. There's not even a way to properly avoid this because distillation is indistinguishable from normal use, the chinese simply subsidize tokens for normal users, and then they just gather usage from these subsidized accounts, while still expensive, this approach is just a fraction of the cost for training. I also suspect that the only way frontier labs are advancing their models by employing a decently sized army of nigerians to do RLHF, which explains the ballooning costs. Every lab downstream from the frontier labs does not need to go through this process, they can comfortably sit back and let Dario do all the heavy lifting, it's ok if they're not leading.

Vibecoding is now good enough to reverse engineer proprietary software to a reasonably good degree. Functional but a bit shitty is how I'd describe current results. Mostly the discourse is focused on people doing it with games and merging them amusing but very much shitty and unstable results, but we're at the point where we can reverse engineer proprietary software in a week or two where it would previously take months or years.
I know there are probably going to be serious negative consequences in the long term, cloud based service models being massively incentivized or draconian IP laws being passed as some examples, but I'm still getting a little bit of schadenfreude seeing this happen to the software industry which is now nothing but a vile mass of spyware, bloatware, and adware.
I once saw a post somewhere where it was pointed that on every level AI is the most anti-IP technology ever made and now we're beginning to see that unfold in bombastic fashion.

>>34503
>draconian IP laws being passed as some examples
it can't be legislated quick enough, and not without pitting some porkies against other porkies because like what kind of features are, like, adobe exclusive, vs what's convention. it's a can of worms.

I think AI decomps are overall a good thing but I'm also amused by the lengths people go to not just pirate things. With stuff like Adobe I guess the copy protection is tougher but most games are super easy to pirate and it's really stupid that people are going "yeah I made a stiff clone of Smash Bros for $x of compute instead of just emulating it for free".

>>34508
People are confusing decomps with recomps

>>34382
Not much suffering since Pipewire came out. And even before then, JACK was probably alright. And I'm fine with the smaller plugin ecosystem, especially since my favourite synth, Vital, is Linux native anyways.
>because sequencing on Ardour used to suck even more
For me, it's alright since I found out how to open the piano roll window. Still a little awkward workflow wise, but eh. https://manual.ardour.org/working-with-midi/pianoroll-window/

>>34384
The suffering was never in the audio server. JACK on alsa plus realtime kernel was always amazing compared to windows. The moment you go for anything under wine whether it's plugins or DAW it's a bad time. The new pianoroll looks good in Ardour, but that wasn't there when I was shopping around for a DAW. I also prefer to have a grid sequencer for my drum machine as well.

https://www.discodsp.com/
Linux native but proprietary

It is amazing how the best softsynth in existence is free and Linux native/open source. You can do basically anything in it. I've actually been using it as basically a subtractive synth, and there are presets that emulate a bunch of historical synth sounds as well. Oh, and what's even more amazing is that the 2nd best softsynth in existence, Surge, is also free and Linux native. https://surge-synthesizer.github.io/manual/

>>34386
>I also prefer to have a grid sequencer for my drum machine as well.
Bitwig has one of those built in IIRC.

https://www.zounoashi.com/#/game/karaage?card=square2

Put any image you want in a deep fryer

ich weiß gar nicht, wie ich das internet in zukunft überhaupt nutzen soll. also wenn wir davon ausgehen müssen, dass langfristig alle kleinen/alternativen foren/diskussionsräume sterben werden wegen gesetzen (kids act), KI und weiteren äähh… externalitäten, dann bleiben nur noch die großen, ultramanipulativen corpo plattformen. ok. ich benutze seit einiger zeit wieder xitter und mein "for you" feed besteht zum größten teil aus titten, schwänzen, muschis und ärschen. was ich ja mag, aber ich kann doch nicht permanent horny sein, wenn ich durch den feed scrolle?? ist das wirklich die zukunft?

>>34288
AI text detection has gotten much, much better. It's an arms race, but at the moment Pangram is so good at detecting AI text that if it flags something as AI it's basically certain that it is.

Somehow all of my search results turned to slopsites or silos.
Imagine you could only search blogs and universities.
If you needed to query a silo, you go to them directly.
https://wiby.me
https://marginalia-search.com/
https://kagi.com/smallweb
https://hister.org/

Recently queried for a fairly obscure topic: "no-mind" and found an interesting page.
Something documenting Chinese philosophy of mind, or lack there of.
The central idea as I read was that heart-mind, and dao fail to distinguish either:

1. a duality between reason and affect/
   
2. a duality between social and personal.
   

Plus there were some later innovations.
https://philosophy.hku.hk/ch/mind.htm

>>34490
Thanks for Marginalia anon, it's awesome for finding things like small blogs and solutions to niche problems

grundrisse the fragment on machines

>>34434
Shit like this is why ID verification for using social media or purchasing electronics is being pushed by the elites, yes it helps expand the surveillance state, at least the ways they're pushing to implement it, but for some it's also a sincere well-meaning attempt to fight back against the dead internet phenomenon.

>>34449
Funny how this website pushes the same thing. They do everything in their power to try and make posting anonymously here impossible for the same reasons as "le elite."

>>34451
>Funny how this website pushes the same thing.

leftypol? how exactly?

These farms existed for years before AI, but in the past they were mostly limited to shitholes like Russia or India. Now that copyright is dead and CFAA in the US is valid only if you're not doing anything related to AI - we're about to witness an explosion of these things.

Captcha evasion through services like 2captcha, phone proxies bought from some shady vendor without KYC laws… That's the thing about dead internet theory. It's not like it was just a theory prior to the invention of LLMs. Mass social media manipulation is something that has existed for nearly 13 years. AI just made it cheaper than sliced bread.

any of these people who take it upon themselves to try to become some kind of extremely-online micro-celebrity armchair journalist crusader constantly seeking conflict and controversy are fucking crazy, normal people who value their sense of self don't act that way and don't feel comfortable reinventing themselves as some kind of superhero character that completely eclipses their actual selves and turns them into some kind of one-dimensional entity rather than a human being.

>>28088
Looks like Every Single Tech company is going to copy Google's pixel 10a design which has an exploding battery and literally sell remote bombs to consumers which could detonate after bitcoin saturated consumers in China reveal their new remote killing system a purposefully made backdoor into the device to kill evil white men.

>>34357
We're SO back.

>>31338
>still talking about antifa clowns in 2026
Dude has a terminal case of boomer brain. Yes, I'm aware he's not literally a boomer.

>>28114
Lets not pretend that every H1B is some best of the best "we can't find someone of this caliber in the US" situation. There is a MASSIVE amount of outright fraud in many overseas tech sector "education" services where someone who just stumbles through a bootcamp is told to say they have 10 years of experience because its very difficult to verify; many in India for instance have "Oh yes, we'll be your reference and say you had a PhD from IIT and 15 years of engineering experience" and, while those applying to the biggest silicon valley companies may get caught, there's both A) a network of existing hires who help others. There was a major scandal a couple years back in Microsoft and another in Oracle I think B) the corporate vested interest not to look too hard because

The real benefit of the H1B is to pay less than a local, legal American AND have someone who is beholden to your company. They can't just work for a few months and then get a better offer elsewhere, their H1B is connected to the particular sponsoring company and if you revoke sponsorship, back they are supposed to go (or remain here illegally, which some do). This is why you see job offerings that have insane qualifications for a low to mid level position, so they can say that there's "nobody qualified here, better go to H1B" but that's not much different from getting illegals to work in agriculture because no American is going to accept $1 per bushel of lettuce picked etc.

So even when they are "educated professionals", they're often being paid a lot less than a local educated professional for the same job with the same ostensible work history. This is not not an issue of "failing to compete / we don't have the workers' any more than big ag can't hire Americans because "nobody wants to work everyone's lazy".

>>34439
Looks sick. Maybe next time.

i've always wanted a pinephone but i have to pay a hefty sum for importing it, the RAM drought also hit it pretty badly and 4GB is just not good enough for a "pro" model. maybe when the chinese flood the market with cheap ram in 2028

>>34439
usecase?

I can't say too much, but if you ever meet somebody of rocketlab, do me a favor and punch them in a coma.

space exploration is like the uncontroversial, bipartisan, reddit-friendly iteration of "make america great again". america had this peak moment of great hope and national pride in the 1960s where astronauts walked on the moon in the most expensive political theater photo op of all time and will never, ever reach a peak like that again because the empire is in its twilight years and everything else in the solar system is hundreds of millions of miles farther away than the moon and just as boring and useless as the moon and there's no reason to go there, but people can't let it go, they so badly want to relive that past glory and to believe that it could happen again and that the naive 1960s optimism about the future space age will actually come to pass. and all the rocket launches are just obscene displays of extravagance and waste, like some trillionaire neo-nazi tech ceo just basically burning billions of dollars in cash on national television while americans watch and applaud like dopes as the entire world turns to shit HAHAHAHA LOOK IT WENT UP OH WOW

>>34415
CHYNA is going to land on the moon in our lifetimes so that's pretty neat

>>34416

they'll probably just build their own moon that costs like $50 and land on that

>>34483
Social media is meant to be a vehicle for talking to people, mainly friends, online. It's not supposed to be cathartic, it's not supposed to be how you get the news or get into arguements with strangers, and it's not meant to be the main way you learn about things like OP is trying to use it for. There's other web technologies for all of those things, you can't just rely on the attention span shredder machine to do it all.

>>34483
It's pretty obvious that anonymous online interaction is different than real life personal interaction so it's weird you ignore that just to post your meme phrase

It's called a passthrough mod. A pretty neat use of depth buffers.
>>34481
The reason why there's a boom of these is because they're not that complicated conceptually, not because AI is doing anything special that you couldn't do manually with a bit of knowledge on modding two or more games.

>>34486
This seems neat on a technical level but I can't imagine it being fun to play.

>>34487
Just seems designed for content slop basically, like everything else nowadays

>>33871
Vanilla android doesn't run anywhere, android is always customized to run anywhere other than in a vm, it's a cursed shitware os. But yes lineageos supports waaay more devices, and the benefit of graphene seems highly questionable to me.
That said all hardware is aids-ridden, google's hardware is not even the worst, probably by far not the worst. Btw assuming you're the guy from the samsung thread - in the past replicant devs found a modem-level backdoor in a samsung phone. And that's just what the found - it's a very small team and the number and complexity of devices is huge and ever growing, so you can safely assume EVERY device has something like that or worse.

looks like GrapheneOS has fixed this now by mirroring the files on their own servers. but I wonder why they do this all the time for every connection rather than say once per hour or once a week
also yes trusting GrapheneOS instead of Google is not without problems but there are degrees in hell

why do people on here argue over what to trust as if anything could possibly be trustworthy.

the smartphone market is just a dumpster fire from top to bottom, particularly android. android is the worst operating system in the history of mankind, the bastard child of google and linux, the two worst things to ever happen to computer technology. degoogling android won't make it better, nothing can make android better, android is shit all the way through, there is literally nothing good about it at all.

This thread glows like the sun.

>>33943
in the final analysis it's all just programs, so build.sh is the trve answer

>>33944
The average build.sh can't do minimal rebuilds, i.e. it will always recompile the entire project, whether you change a line in config.h or not. Redo targets are literally just executable files, that specify dependencies through redo-* calls, so repeated redo invocations will only rebuild targets according to the dependency graph, while those targets are still arbitrary programs.

>>33945
just buy more CPU time

>Sudo apt upgrade on kubuntu system
<kde all broken

ubuntu is by far and large the fucking worst distro I ever had to touch, they deserve to go bankrupt

>>34418
Fixed, guilty party was chromium, removed it and everything works again https://issues.chromium.org/issues/565132857

Am I wrong in wondering how the fuck a glaring bug like this gets pushed through? How did they not notice this? Is this because they just don't even bother anymore with snap?