Live data from Hacker News

Saying Goodbye to GitHub

ersei.net

321–330 of 450 posts

Re: Saying Goodbye to GitHub

#321
post #316

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'…

Perhaps I'm out of the loop on this, but I always thought the concept of open source was primarily about the opportunity for personal professional development. The ability for someone not connected with a corporation to stay relevant and continuously update his skills in away that was not dependent on proprietary systems. That is a huge asset, not only for oneself but also for the the world.

That may be a benefit, but the primary concept of open source is that its open source.

Re: Saying Goodbye to GitHub

#322
post #98

Earlier quoted context omitted.

Not given back to the original authors, but given to the users.

Only if you give those users your modified version. If I host GPL software on a webserver and my users use that webserver, I don’t have to give them the source code for modified GPL programs. This is fairly common.

Depends on which GPL you’re using.

Re: Saying Goodbye to GitHub

#323

Earlier quoted context omitted.

I agree that AI usage of code is somewhat murky with current licenses,which obviously don't mention it either way. Free software has a principle of "freedom to run, to do whatever you wish" (freedom 0), so arguably has said that training AI is OK. (We could quibble over the word Run, but the Gnu.org,and RMS clearly say "freedom 0 does not restrict how you use it." GPL code can be used by the military to develop nucle…

I largely agree with you, but I think there is one question that hasn't been addressed yet: Are the weights learned by an LLM a derivative work? When a person learns from GPL code this question doesn't arise. The state of a person's brain is outside of copyright. But is the state of an LLM also outside of copyright or outside of the terms covered by the GPL? I'm not sure. An LLM can not only emit source code derived…

> the state of a person's brain is outside of copyright

It clearly isn’t. Which is why clean-room reverse engineering always requires at least two people. Or why a musician that accidentally recreates a chord progression they heard years ago but don't remember the source might still get sued.

Re: Saying Goodbye to GitHub

#324
The main problem is that GitHub is becoming unreliable like the rest of the MS software (lead by Azure).

Everything else is just business.

At the same time I don't think it's smart to abandon GitHub for networking / CV reasons.

Once I'll be rich enough, I'll delete my account and use a self hosted git server.

Re: Saying Goodbye to GitHub

#325
well, GitHub isn't as "raw" as it used to be, that's for sure. It's much more user friendly now, much easier to make a single click and do what you need, like "copy to clipboard" or "download (a single file from a repo)" and so on.

as things get easier to use, the coolness factor definitely goes down, and the pride you get from learning the tool goes down, because as things get easier to learn, the less special any given person is for learning that tool. There are software packages which intentionally remain complex to keep this feeling, and it's considered bad.

From personal experience, if you want GitHub to feel like magic again, host an enterprise server instance on AWS and have over 10k active users. You will feel the complexity in your soul as you work out ways to monitor that thing effectively and react to problems before the symptoms get noticed by your extremely demanding users.

I realize that very few people can do this, as GitHub Enterprise licenses are expensive, but hoo boy GitHub has not lost its magic for me.

Re: Saying Goodbye to GitHub

#326
> For just $10 a month or $100 a year, you too can get programming help with no obligation to give back to anyone!

I would say the use of trained data is "giving back" to all users of Copilot?

The only thing that has to be considered is if the pricing is fair, and with all the competition in the space will be soon, or may already be below cost.

Also in my opinion, not using open-source for programming aids is less freedom (libre)--without getting into specifics of compatibility of doing so with individual licenses.

Re: Saying Goodbye to GitHub

#327
post #75

Earlier quoted context omitted.

I just want to make sure you appreciate that if you really believe this argument then GitHub can only be used by the people who actually directly own the copyright on projects; and if you, for example, want to clone and edit my software (the vast majority of which I explicitly never uploaded to GitHub) then you wouldn't be allowed to (which doesn't seem like either the intention or the way it is commonly used)... and…

I'm pretty sure thats a narrower interpretation than GitHub are aiming for. I'm just paraphrasing the parts of GitHub's ToS that I can remember since current debate on the topic has lead to me remembering a few important parts reasonably well but I've certainly not memorised them. So this is a good opportunity for me to go re-read them and quote them directly... (also in case anyone is about to mention it ... I am aw…

The key distinction I'm interested in is providing the GitHub (or any similar product) "Service" vs selling a separate, derived product (Copilot / ChatGPT).

A: Common ToS to say that a product's owner obtains a license to user content for purposes of providing that user the product service.

B: Somewhat common ToS to extend that to providing the product service to third party users (i.e. use your content for other users of the service), but depends on business model (e.g. most social* businesses).

C: A lot less common ToS to obtain a right to distribute user content in derived products.

A number of sites have gotten into hot water with their userbase over trying to update their ToS from B to C. From memory... Adobe Cloud, DeviantArt, maybe some others?

Typically this gets flak in creative communities, given that it is many people's business, and they're more concerned about distribution rights than your average coder.

At its base, OpenAI/Microsoft/etc. will eventually run into the exact same issues that bedeviled the Linux kernel in the 1990s, except with a much thornier IP ownership question (given the greater number of parties).

Re: Saying Goodbye to GitHub

#328

Earlier quoted context omitted.

> Leaving GitHub wont change that, OpenAI is training its models on every bit of code they can have, sourcehut, codeberg etc. If its public, they will train on it. Not every bit of code, they are respecting proprietary licenses. When MS puts the code for Windows, Office, Azure and everything else in front of ChatGPT, Copilot, whatever other AI learning model they have, then perhaps they have a leg to stand on. Otherw…

AFAIK that has nothing to do with the license, it has to do with whether the code is public. You don't want the AI accidentally revealing proprietary non-public information (e.g. imagine someone had a secret API key in a private repo and copilot leaked it; that'd be a huge incident), so you don't train it on that information, regardless of what it's licensed under. You could make a similar argument for not training o…

> Not to mention it's way easier to delete an accidentally-added snippet of GPL code from a codebase than it is to "unleak" company secrets after they've been publicly revealed.

How often do you think anyone will notice that some part of a proprietary codebase is copied substantially from GPL code? I think it's going to be very rare and a lot of this code will fly under the radar. The GPL was always a kind of legal jiu-jitsu, turning copyright against itself and allowing non-commercial entities to protect themselves from uncompensated exploitation. Models like copilot, if they're legal, upend the status quo tremendously. Even though your code isn't (always) used directly, a commercial entity like Microsoft will slurp it up and sell the resulting model back to you for $9.99/mo.

Re: Saying Goodbye to GitHub

#329

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'…

I wonder if the GPLv4 will be coming out soon with an anti-AI-training clause (unless your output model and its outputs are also GPLv4, of course).

Re: Saying Goodbye to GitHub

#330

I kept my github account to report issues on the projects still on github. Even though github is own by the untrustable msft, it (still) does work with noscript/basic (x)html browsers (well I created my account probably a decade ago). gitlab: I cannot even create an account.

Stop reporting issues to github.

Just had to report a bug on the alsa-lib... and it has to be done on github.

Ok, we have a situation here. Critical/significant open source components going on sites which should be avoided. There are noscript/basic (x)html alternatives.

Why those are not chosen??

Post reply on HN