Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Saying Goodbye to GitHub

ersei.net

Post reply on HN
Print view View original

371–380 of 450 posts

Re: Saying Goodbye to GitHub

#371

Post by welshwelsh »

lenkite wrote: post #258
bruce511 wrote: post #11

I'm gonna get hated on for this, but I don't think "give back" is an open source concept. I'm not aware of any Open Source license,or Free license for that matter,that has a give-back clause. Source code is available to -users- ,not prior-authors. Some Open Source licenses can be used in proprietary code, (MIT, BSD etc) with little more than simple attribution. Those developers chose that license for a reason, and I'…

IMHO, just like there was a robots.txt file made for the web, there needs to be a NOAI.txt for git repos. Sorry, this repo does not permit you to ingest the code for a learning model. Seems completely reasonable.

What would be the point of that?

If we were somehow able to prevent AI models from ingesting a codebase, that would mean everyone else who wants to produce similar code would have to re-invent the wheel, wasting their time repeating work that has already been done.

All because... the person who did it first wants attribution? They want their name to be included in some credits.txt file that nobody will ever read? That's ridiculous.

Re: Saying Goodbye to GitHub

#372

Post by marcosdumay »

gavinhoward wrote: post #357
marcosdumay wrote: post #351

Earlier quoted context omitted.

I'm out of the loop. What drama has been with Gitea? (I'm not a bit willing to put my code on the likes of SourceHut or Codeberg. They are just new iterations on the Sourceforge/GitHub pattern that didn't go bad up to now . Besides, I like to keep the software I don't publish on my computers.)

See https://gitea-open-letter.coding.social/ . It should have links to more. Comments at https://news.ycombinator.com/item?id=33372471 .

Oh, interesting.

The silver lining seems to be that it forced the creation of a fork, promised to stay free, with actual documentation, and that doesn't force the docker bullshit upon installation.

Looks like I'll finally upgrade from Gogs.

Re: Saying Goodbye to GitHub

#373

Post by EthanHeilman »

b3morales wrote: post #356
EthanHeilman wrote: post #346

Earlier quoted context omitted.

It would be interesting to have Free Software License that requires that any thing which ingests the source code must be Free Software running on Free Hardware. If you train a model on such inputs, your model would need to be Free software and all the hardware the model runs on would need to be Free Hardware. This would create a massive incentivize to either not use such software in your model or to use Free Software…

This idea ("make a stronger license") has come up in previous discussions of Copilot as well[0]. The problem is that the Copilot project doesn't claim to be abiding by the license(s) of the ingested code. The reply to licensing concerns was that licensing doesn't apply to their use. So unfortunately they would just claim they could ignore your hypothetical Free³ license as well. [0]: https://news.ycombinator.com/item…

> The problem is that the Copilot project doesn't claim to be abiding by the license(s) of the ingested code. The reply to licensing concerns was that licensing doesn't apply to their use.

I think github is largely correct in their view on licenses. However I would argue that you could create a stronger legal binding than say a GPL-3 license. For instance you could require and enforce that anyone that wishes to read the repo must sign a legal contract or EULA: "By decrypting this git repo you are agreeing to the following licenses, restrictions, contractual obligations, ..."

Re: Saying Goodbye to GitHub

#374

Post by jgerrish »

FrustratedMonky wrote: post #311
bombolo wrote: post #215

Earlier quoted context omitted.

Would I be able to train an AI only using microsoft's leaked windows code to write a windows clone with no copyright (since it comes from an AI) and be safe from legal repercussions because it was trained on fair use code I just happened to find online? If they rule that it's ok to do that, I might be ok with AI being ruled as fair use.

Exactly. I'm betting if you asked GPT to create a windows clone, for sure MS would not let you distribute. This will go like every other law/license, big corp can sue little guy into the ground. When big corp uses your code it will be 'thats just a model generated code not yours'. But in other direction, if little guys creates windows clone, 'sorry, its off to jail for you maytee'.

Even if it's the opposite direction, big guy losing and small guy coming out ahead, it's still drama.

Just like Covid introduced epidemiological terms to the general public, this issue can introduce design choices around licensing, copyright and watermarking to more people.

I assume there is a group of researchers building tools to provide fine-grained historical views into AI output. And yes, for billions of parameters trained on billions of documents, linking every letter to a source document is a UX nightmare.

But what a cool problem. That's the interesting part. Yeah, something like TileBars[1] or Seesoft[1] seems like the right tool. But maybe keeping it all text with some graphical marker of authenticity is the better choice.

So many cool problems. But, that authenticity marker is the hard sell. Can reasoned discussions with others be enough to introduce that, or is drama required?

https://people.ischool.berkeley.edu/~hearst/irbook/10/node7....

Re: Saying Goodbye to GitHub

#375

Post by xigoi »

jacquesm wrote: post #10
aetherspawn wrote: post #4

Why hasn't someone just changed the GPL license already: "If you train an AI on this code, you must release the source code and generated neural net of that AI as open source" or something to that effect. It won't stop it, but it will slow it down, and it seems like the right T&Cs to put on training against GPL code because it gives an advantage to open source AIs, however minor.

They are really not going to care about what you put into your license file, they are just going to claim that the use of GitHub binds you to their terms of service and that this supersedes your own license. Good luck fighting that.

> This license does not grant GitHub the right to sell Your Content. It also does not grant GitHub the right to otherwise distribute or use Your Content outside of our provision of the Service, except that as part of the right to archive Your Content, GitHub may permit our partners to store and archive Your Content in public repositories in connection with the GitHub Arctic Code Vault and GitHub Archive Program.

Re: Saying Goodbye to GitHub

#376

Post by xigoi »

swyx wrote: post #49
npinsker wrote: post #19

Earlier quoted context omitted.

I appreciated the specific word choice there; 'disgusting' implies a strong, visceral reaction, as if to say GitLab's work is so far from the author's taste that it blocks them from considering the product, and didn't seem all that harsh if that's how the author actually feels. In a counterintuitive way, it was nice to read -- it says that UI is important; that it shouldn't be an afterthought; that it has the power t…

what is so bad about the UI/UX? i'm not a frequent gitlab user but it doesnt strike me as anything out of the ordinary

The main page has a huge prominent “Get Free Trial” button, while the “Sign In” button is small and hidden in the menu. A common dark pattern.

The performance is sluggish.

There's always a lot of buttons and stuff on the screen that most users will probably never use, making the UI cluttered.

Re: Saying Goodbye to GitHub

#377

Post by theonlybutlet »

EarlKing wrote: post #359
theonlybutlet wrote: post #295

Earlier quoted context omitted.

That's not the case, there's a probability it may "spit back" the protected expression. There's also a probability I, as a human "spit back" the protected expressions. This could either be by pure chance or from past learnings, reading the protected code and internalizing it as a solution, my subconscious forgetting I actually saw it elsewhere. In Uni, students run their theses through plagiarism checkers, even if it…

...except you don't need an infinite number of monkeys. It has been trained to produce protected expressions by virtue of being trained on protected expressions. The probability of it producing a protected expression at some point is 1.

The same truth holds for you or me writing up that code.

Re: Saying Goodbye to GitHub

#378

Post by WorldMaker »

Uvix wrote: post #278
gampleman wrote: post #255

Earlier quoted context omitted.

From the MIT license: > The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. Does GPT spit out the copyright notice when it regurgitates my code?

Is GPT spitting out “substantial portions” of your code?

It might. The ethical dilemma here seems to be:

1) We don't know when/if GPT will. GPT in its current form can't seem to guarantee safety (either from "substantial" verbatim snippets or from complex "hallucinations" of random pachinko output).

2) GPT doesn't know when/if it has. GPT in its current form likely cannot know this. (In part, because it doesn't really "know" anything, that's too anthropological a word for what is still mostly just a casino full of pachinko machines.)

3) Define "substantial portions" in a way that a jury of your peers can understand it in a court of law.

4) Can you define "substantial portions" again, only this time in code as guide rails for something like GPT? "Substantial portions" is barely a human term designed for human lawyers and courts. There's a fascinating challenge here on quantifying it.

Re: Saying Goodbye to GitHub

#379

Post by mftrhu »

welshwelsh wrote: post #371
lenkite wrote: post #258

Earlier quoted context omitted.

IMHO, just like there was a robots.txt file made for the web, there needs to be a NOAI.txt for git repos. Sorry, this repo does not permit you to ingest the code for a learning model. Seems completely reasonable.

What would be the point of that? If we were somehow able to prevent AI models from ingesting a codebase, that would mean everyone else who wants to produce similar code would have to re-invent the wheel, wasting their time repeating work that has already been done. All because... the person who did it first wants attribution? They want their name to be included in some credits.txt file that nobody will ever read? Tha…

> All because... the person who did it first wants attribution? They want their name to be included in some credits.txt file that nobody will ever read?

Yes, and yes. Those would be the terms that person publishes their code under. If you can't agree to those terms - maybe because including a single name in a credits.txt file that no-one reads is somehow too onerous for your process - then you are always free to re-implement that code on your own.

Re: Saying Goodbye to GitHub

#380

Post by mftrhu »

HenriTEL wrote: post #177
bruce511 wrote: post #11

I'm gonna get hated on for this, but I don't think "give back" is an open source concept. I'm not aware of any Open Source license,or Free license for that matter,that has a give-back clause. Source code is available to -users- ,not prior-authors. Some Open Source licenses can be used in proprietary code, (MIT, BSD etc) with little more than simple attribution. Those developers chose that license for a reason, and I'…

> On the one hand we happily train humans on GPL code. Those humans can then write their own functions,but for trivial functions they're gonna look a lot like GPL Source. Exactly. People are getting mad that Microsoft is making good money while the people who made all that free software available mostly did it for free (like in no money and no recognition). It can sound unfair but that's the deal. If you didn't want…

> If you didn't want people or AI to learn from your code, open source was not the right option.

There's nothing wrong with other people using - learning and creating derivative works of - one's open-source code, provided they respect the terms of the license. It seems to me that the real issue is the fact that these licenses don't have enough teeth.

Post reply on HN