@hopeless@mas.to

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@tante Nice to agree about something... I don't need integrated AI features (or Ubuntu) either. Stock Debian is good, Rocky (FOSS rebuild of RHEL) is also good with overlapping 10y lifespans. https://medium.com/@bokiko/ubuntu-is-free-but-the-company-behind-it-made-292-million-last-year-213c3ab5351a

medium.com

2

3

0

degenerating degenerate @hopeless@mas.to

· 4mo ago

Replying to

@jik@federate.social If you just issue terse toplevel demands, they do the minimum needed to tick your box and the results are not maintainable or good. But... 1) if you "do the engineering" in the design and ask it to work just on each phase in turn, which is maintainable, they will work within your considered constraints. So, you design the bridge and outsource the parts to the foundries. 2) If it only lives as long as a Google product, or less, does it actually need to score 100% on maintainability?

1

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@gabrielesvelto @fabrice ... watch a streaming service like Netfix? Every time you or anyone else watches a few 4K movies, that's 4GB downloaded. Every night, the streamers have hundreds of millions of users. For better or worse mass distribution of 4GB is just "Tuesday" for Google in 2026 with or without the local model aspect.

1

9

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@tante You actually ARE extra capable using a coding assist, or finding vulns via AI... it's not an imaginary or unmeasureable state. And AFAIK it's having zero effect on my lungs. Still I agree the current way things are, where there's no equivalent self-hostable system right now, it's a serious problem. It's not simply it's essentially being forced to use proprietary systems, it's also that those systems are under US control, run on your actual (FOSS or not) machine over the internet.

1

6

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@bjb @tomgag Depends what you mean by 'compatible'... in the sense of AI holding copyright in its output like a human, it seems it's not going to play out like that. In the sense of a human owning the copyright on AI output, sure. A photographer definitely owns the copyright on images his camera took, and he essentially just pointed it in a direction and pressed a button. I don't think copyright + AI coding assists is going to be any special problem.

1

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@evgandr @mrmasterkeyboard @chikl What an unfortunate strawman forced to work on FOSS "or die in poverty" by these nasty bosses you invented. If you maintain a reasonably sized FOSS project you will soon find lots of not enjoyable subsidiary tasks that must be done. And aside from those are the users, with literally thousands of complaints and demands. It might have started with "intellectual puzzles" but it will inevitably become unpaid, unthanked grind. Coding assists are a godsend.

0

2

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Oh an unexpectedly great article via lobste.rs

https://blog.feld.me/posts/2026/04/open-source-does-not-imply-open-community/

> Github turned all of open source into an unpaid job for maintainers.

> Suddenly open source is a second job. You're burned out.

Yeah. And your "community" feels legitimized by it, you get explained how just being a user is a contribution.

Makefile.feld

Open Source Does Not Imply Open Community

Open source software has existed long before the invention of the (D)VCS. The author likely hosted a barebones HTML webpage or a txt file describing the project. There definitely was an FTP server somewhere with tarballs. The author may have been reachable by email. If you were really lucky …

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@lispi314@udongein.xyz @lina@vt.social Yeah. I don't know why there is such a Copyright Maximalism speedrun going on here.

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@tomgag I did not read about any ongoing legal action about the trademark. Instead both sides mentioned in the story are going to try to fight it out taking the project their own way. I explained already whoever can invest more effort typically ends up driving the project. And that the guy who uses AI will be able to invest more effort than the team with more people, who exclude using AI. If you have a different take, fine; it should be easy to show what's wrong with my points?

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@gabrielesvelto @fabrice Well... for you and me if we get 4GB we only see 4GB coming. For Google they see n x 4GB going out, where n is in the billions. So their motivation to compress it will be magnified. Fedora also has the same asymmetric situation around updates, and for many years did binary delta packages in the repo between recent -> new versions. They gave it up at F40, there were gains (of 8%) but if the delta failed to apply you ended up needing 192% https://fedoraproject.org/wiki/Changes/Drop_Delta_RPMs

fedoraproject.org

Changes/Drop Delta RPMs - Fedora Project Wiki

0

2

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@lina@vt.social @Plettigoal@sportsbots.xyz If you imagine yourself as Gordon Gecko looking for an easy meal, he still has to point to some of your code and say, "Your Honour, this for loop is almost identical to the for loop in my beloved SCO Linux" only to get a chart in the courtroom from the defence showing 100,000 examples of the same structure from prior art gleaned from FOSS on github. Unless it's verbatim, which modern coding assists never are, they cannot even show any supposed lineage from The Precious.

0

2

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@lina@vt.social @florian@owo.ff15.eu

No it is exactly how it works.

"... the copyright owner’s actual damages and any additional profits of the infringer..."

Actual damages for works under MIT: $0. Additional profits when your own work is FOSS: $0.

"... statutory damages..."

"...to be eligible for statutory damages and attorneys_fees, the creator must have registered their work with the u.s._copyright_office before the infringement occurred (or within three months of publication)."

https://uslawexplained.com/17_usc_504

uslawexplained.com

17 U.S.C. § 504: The Ultimate Guide to Copyright Damages and Profits

Learn how 17 U.S.C. § 504 governs copyright damages and profits, including statutory damages up to $150,000 per infringed work, and what it means for you.

0

5

0

degenerating degenerate @hopeless@mas.to

· 4mo ago

Surprise of the day... the excellent EU push to force laptop makers to move to USB PD has, after 18h with a new laptop, made all other laptops with proprietary PSUs seem like relics from some other era, like rotary-dial telephones.

It's surprising because my first USB-PD laptop was from 2019 or so. Somehow the first one after knowing they'll all be now like that has ushered in a new psychological era.

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@scopecreeppress @tomgag If you are looking after the code and forcing the incremental drops to be maintainable iteratively, AI code can be fine even in Apr 2026.

0

1

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@lina@vt.social @florian@owo.ff15.eu I'm glad you're making money, but my gross revenue is $0. It is very easy for me to show $0 profit on my FOSS work that is part-written by coding assists. And I remind you, huggingface would be the first victim if any of this was relevant. Also the point... they had to register their works at the US Copyright Office within 3 months to "get nukes". I guess some people like MSFT did this, but people who put their sources on github to be scraped under a liberal license?

0

3

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@tomgag ... case in point is scopecreeppress, who sprayed his opinion here as if it especially mattered and then blocked me presumably for simply politely disagreeing with him. Instead of arguing and defending (or heaven forbid, modifying) his point and maybe we all learn something, his claims are vomited out and left like a bad smell, and then he's gone, forbidding any reply and guaranteeing the last, useless word and turning off the pain receptors.

0

2

0

degenerating degenerate @hopeless@mas.to

· 4mo ago

Replying to

@stefan_hessbrueggen@fedihum.org @theodoraward@mastodon.social @tante@tldr.nettime.org Yeah... so let's say for argument purposes it's proven as a fact they are "not conscious" and say we all agree on it. What did that change? They are still capable and can reason, and "think" to achieve goals. It's completely meaningless, same way 50 years ago the medical profession insisting animals "do not feel pain" and "are not conscious" did not change the situation for animals that can also reason and feel pain. I can note one thing though, LLMs have won a lot of arguments in the last year if this is currently the frontline for the resistance!

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@lina @Kensan (Related to previous points) Yes... but... they have some things in common with people... you can discuss and explain things with them, and at least in the non-user model training mode, they can "learn", they can "read". "take" and "steal" are not copyright things, if I read a book, the book is still there and someone else can read it. Copyright infringement is not the same as "stealing". If "Big AI" happen to want something that is true, it doesn't mean it's wrong suddenly.

0

2

0

degenerating degenerate @hopeless@mas.to

· 4mo ago

Antigravity added a lot of interactive command auth stuff... really this is security theatre since ultimately the whole goal is to let the assist edit code, build and run it. The theatrics include not letting you whitelist kill or killall... well I can see it... but... the thing writes code. It can just call the same apis as kill[all].

They also added a 'sandbox', but it's incompatible with valgrind.

It feels like being funneled into turning off the auth which would be even worse.

0

degenerating degenerate @hopeless@mas.to

· 4mo ago

Replying to

@tante@tldr.nettime.org It's a good article... but doesn't it feel like the same article that would have been written during the dot com boom / internet bubble? It seems mainly about work - life balance or lack of it...

0

2

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@inquiline@assemblag.es Being harrassed certainly is bad for anyone targeted. But also a 'social media' that is essentially unchallengeable 'write only' slogans and chants is not useful... ultimately even to the author who wouldn't've considered the opposing points.

0

7

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@mtnrbq65 @ekis Well, we are talking in OP's cope / denial thread. That's also "important context" for how you can expect to be read here. I heard a section about exactly this and the AISI report on Today on Radio 4... they had two guys on who interviewed very well about it. The main thrust was it's not really about Anthropic or Mythos, the gap between creating the zero-day, the ease of finding it, and deploying it has basically evaporated due to AI. The buzzword was "machine speed".

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@mrmasterkeyboard @chikl So you "called out" this Free Software that you presumably use... what are you going to say to them exactly? "I am deeply troubled that you guys might have more time with your families or sleeping if you use coding assists!", or "Hell no I am not going to write that code manually and contribute it. I just want to see you suffering doing it manually." "Somebody used AI to review a contribution they didn't even ask for? String 'em up!"

0

11

0

degenerating degenerate @hopeless@mas.to

· 4mo ago

Replying to

@jenniferplusplus@hachyderm.io Have you actually tried current-gen coding agents like https://antigravity.google What you wrote sounds like you should have tried them. But what you wrote also sounds like you don't know what you're talking about.

Google Antigravity

Experience liftoff with the next-gen agent platform

0

4

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@marmeladenbaguette @tante I took your comment about the lung situation seriously, it is a reasonable point. Please help me understand how you would reply to stranger demanding that you "prove" something to them, that you think would "show interest in [their] replies"? Just say, "no"?

0

1

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@mrmasterkeyboard @chikl

Don’t argue with me over this.

"AI is fascist tech" "fascist companies" "stealing" "unethical" "immoral"

If you want to argue, you have to produce argumentation, not slogans.

Github is mainly full of liberally-licensed software (that's why the sources are just sitting there) with licenses that don't disallow AI use for learning.

Anyway have a nice day.

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

https://noahbogart.com/posts/2026-05-03-losing-skills/

People are using coding AIs and correctly noticing their day is not spent the same way... it's similar to when people "entered management" and so stopped doing output-work, and started directing others to do it.

This guy likens it to using lane-assist in his car. Manually minimizing lane drift WAS a critical skill... after relying on the assist, his manual skill atrophied.

OK... but... if every car has lane-assist, the value of doing it well manually is now zero ...?

noahbogart.com

Losing Skills

For the last year, I've been driving a 2025 Toyota Sienna, a boat of a minivan. It's got all sorts of fancy features that I won't get into right now, but the big one that bothers me the most is a "lane drift" sensor that makes the steering wheel vibrate if you start to leave your lane (without a turn signal) and will actually turn the wheel/redirect the vehicle for you if you drift too far. Ostensibly, this is a great feature, as it helps you notice a Very Bad Thing, but I've found that it's had

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@MildDrop72@masto.ai As I mentioned, if I compose a MIT project in my own MIT project, I am very careful to keep their attribution on those parts.

But when that project was learning material among hundreds of thousands of others and I get a coding assist to write new code, to my specifications, what does that have to do with that one MIT drop in the ocean? Please, go ahead and show the judge.

MIT ... Lost revenue ... other license

Maybe, for GPL3: not a free license that already gives it all away.

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@reiddragon @chikl @mrmasterkeyboard People who aren't too far down the spectrum can decide when to call it a day, especially when it's unpaid. If you take that approach, coding assists will generally get whatever you were trying to do, done quicker. It's up to you what you do with that extra time, keep doing more, or go touch some grass. When you eventually fall ill, as we all will, you won't be wishing you had done more work with your time.

0

2

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@reiddragon @chikl @mrmasterkeyboard I do understand about the pressure be it issues or "community" but at some point you reached 100%. LLMs produce gaps and bugs different from humans and it eats time. But they also allow addressing problems humans lack motivation for, so you can do "stuff that has been [historically] impossible" for you. Debugging via LLM can be a spiral, but you can notice it has been YOLO-guessing and eg, force it to log and find evidence to converge on the problem.

0

degenerating degenerate @hopeless@mas.to

· 4mo ago

It's interesting that the ways the training incorporates a lot of US cultural expectations. They are relentlessly upbeat about whatever you wanted to achieve. They won't argue with the boss although they will point out inconsistencies with what they're working on. It's only a matter of time before they carry guns.

Since most of us are from different cultures, and the US had led most advances since before WW2, could this virtual cross-pollination actually be psychologically advantageous?

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@lina @Kensan "Take" is doing a lot of work here... with LLMs as coding assists, they don't wholesale regurgitate their training content in my experience (with Gemini / Antigravity). If the content is publicly available on the web, it's fair to imagine the copyright holder is okay with people reading / discussing / remembering / learning from it... readers made a temporary local copy so they could internalize it. Why's it so out of the question to frame LLM use of it like that?

0

7

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@lina@vt.social

Yes, this means that anyone downloading "open" models potentially puts themselves in as much legal risk as torrenting a movie does.

Huggingface still seems to be operating?

Also there is a big difference between bidirectionally torrenting somethng and "downloading" it... they are not interchangeable terms; normal humans are not getting into legal peril for downloading / streaming movies or music things (because it is not their "performance" of it but whoever is sending it to them).

0

3

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@mtnrbq65 @ekis "Important context" != selective quoting. The same document also says: "As a first step towards measuring this, we built "The Last Ones" (TLO): a 32-step corporate network attack simulation ... ...Claude Mythos Preview is the first model to solve TLO from start to finish, in 3 out of its 10 attempts. Across all its attempts, the model completed an average of 22 out of 32 steps. Claude Opus 4.6 is the next best performing model and completed an average of 16 steps." See fig3

0

2

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@lina @Kensan Sure. It is very complicated in practice. A judge or a jury will probably upend things multiple times before it's settled. I just would like to see that reflected in the very-normalized-on-masto claims like "unethical", "immoral", "stealing" and so on as if they are facts. For derivative works, the way the coding assists work, it only seems realistic if asked to produce something identical, same language, no other constraints etc. Anyway I genuinely hope you have a nice day.

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@lina @Kensan (Related to new points) Copyright law only does what it does... in countries like the US there is also "fair use" exceptions. The Berne Convention is the basis of modern worldwide copyright https://en.wikipedia.org/wiki/Berne_Convention There are layers upon layers and differences by country but it centrally concerns itself with "reproduction" of the works. https://en.wikipedia.org/wiki/Berne_three-step_test Having used the coding assists for some time now I am pretty sure it does not "reproduce" the works it was trained on.

en.wikipedia.org

0

2

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@yiningkarlli I just want to ask the author of the quoted X post... who has quite a cursed name... WTF happens after a "decade" he mentions in the last line?

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@marmeladenbaguette @tante

its affecting someones lungs

Yes it's not good. I would much rather run local LLM from solar like everything else in my house. But I can't find a competitive local solution to Gemini 3.1 Pro.

also i have yet to see some proof

I didn't get the memo that anything I do depends on your opinion? Please bear in mind I never heard of you until a moment ago.

You're welcome to keep coping with the idea coding assists don't work and I am just an inveterate liar.

0

3

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@inquiline@assemblag.es Spam replies isn't much use. But do you really see that here? Maybe I just live a sheltered life. A year or two ago the batch of refugees from twitter would reply to any reply not gripping their arm and shaking their hand with "reply guy". That's a great way to completely kill discourse too. It's those guys I think about with "challenged". They do not want a better understanding. They just want their write-only opinion feted. Abuse tools shouldn't enable that by the back door.

0

5

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@inquiline@assemblag.es Yes I did take a look at your link initially. It's to put the counter-point to the assumptions in that I replied. I'm aware this is not entirely welcome discussion and after this I'll leave the last word to you. IMHO posters should not generally have control over the speech of others. That there's a reply box is part of the rules of engagement here when deciding to post or not. Silencing others (with hostility...) can be another type of abuse. Anyway, have a nice evening.

0

1

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Today's lobste.rs pulse monitoring: https://larsfaye.com/articles/agentic-coding-is-a-trap

So that sounds dire, but actually the author is a quite complete coding assist believer and user. He discusses various reasonable (ie, based in fact, if not complete prescriptions) problems and outcomes from using them.

At the end he says he uses them heavily, but his preference to fight the slop is to write 20%+ of the code manually and have them work filling that in.

Well, OK. As he says, his quality went up, but not his speed.

larsfaye.com

Agentic Coding is a Trap | Lars Faye

Remaining vigilant about cognitive debt and atrophy.

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@13xforever@social.treehouse.systems @paul@tapbots.social I guess it's both... a) you have to load the unusual crypto module to get zapped, and b) all the distros ship with kernels that have the vulnerable module that would allow you to get zapped I have to say it seems like only yesterday some genius was writing on Masto that only "idiots" would use AI for finding vulns... and then this (found by AI) popped up... oh it actually was only yesterday

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@lina@vt.social @florian@owo.ff15.eu My point is that to attack someone through the courts, you must have standing - it's your copyright - and be able to show damages. The people who licensed their work under MIT are already at peace with $0 "Permission is hereby granted, free of charge, to any person obtaining a copy of this software... to deal in the Software without restriction, including without limitation the rights .." What damages would they show? Copyright maximalism is not compatible with FOSS.

0

9

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@baldur@toot.cafe You'll be happy to hear AIs definitely degrade code for the same reasons, one context only has a feeble grip on the thinking of the previous context. And as they "age" (ie, their context fills up) they get very anxious to just finish exactly what you asked without care. You can manage it and get good results, but you are constantly having to watch it, clean up, and intercede forcing it to redo things to be maintainable. Yet, it's still way faster than doing it all manually...

0

degenerating degenerate @hopeless@mas.to

· 4mo ago

Replying to

@oscherler@tooting.ch @worik@mastodon.social

because the model is a useless piece of junk

I often wonder when I read this kind of macho assertion, what's going to happen when the writer encounters evidence in their life that they are perhaps not entirely correct about things?

If OP meant how the model is managed to iterate at one level or another is critical than it seems that so long as the model is "good enough", that feels like it could easily be.

0

degenerating degenerate @hopeless@mas.to

· 4mo ago

What on earth has happened at Coverity?

Coverity Scan is down

We're performing some upgrades to the Scan servers. During this time, login is disabled. All changes to projects and accounts are disabled.

We're really sorry for the inconvenience. We're working quickly to be back online soon!

https://scan.coverity.com/

I sent them an email on the 20th May asking when they would be coming back up, I just got notified it was "undelivered" due to "Connection refused", I have emailed them fine before.

scan.coverity.com

Coverity Scan - Static Analysis

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@lina@vt.social @Plettigoal@sportsbots.xyz Just to be clear licenses have some power only because they can provide additional grants to negate the underlying law. If the law doesn't provide penalties for the situation, eg, fair use, or your situation isn't applicable to the law, the license is simply impotent. You're then not "breaking" any license; most of my stuff is MIT and I follow any requirements to compose other compatible projects. Anyway have a nice day.

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@fabrice @gabrielesvelto Yeah... even my cheapo mobile plan comes with 30GB/mo transfer... so even if that was all the bandwidth I had, it's only 4 days-worth of it. And I also have unmetered gbit fibre; that's 4GB every 40s. There are other scary points about what Google did and how they did it, but the bandwidth aspect in 2026 is just not a big deal. It feels like if it's related to AI (even on-prem!), any stick will do to beat it to death.

0

4

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@lina@vt.social @florian@owo.ff15.eu Who has standing and losses to legally attack anyone for this? Say I didn't follow these (arbitrary...) demands and I downloaded a MIT project from github as anyone can, and I put it into an LLM. The maintainer of the MIT project has enough mental flexibility to look up from the concise, liberal text of the MIT license, and formulate a complaint to the judge about his losses and ownership of the (completely different) code the LLM emitted? I guess we will all see, right?

0

11

0

degenerating degenerate @hopeless@mas.to

· 5mo ago

Replying to

@evgandr @mrmasterkeyboard @chikl

No I mean a reasonably-sized FOSS project... there are a lot of them in github.

Lots of things are a grind, talking to human LLMs on the internet for example, who can generate output on any topic without needing any initial parameters.

For people without natural intelligence?

Gosh the reality that there's a good, ethical use for LLMs just brings out the worst in people doesn't it.

You have a nice day.

0