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

Give up GitHub: The time has come

sfconservancy.org

Post reply on HN
Print view View original

461–470 of 585 posts

Re: Give up GitHub: The time has come

#461

Post by janosdebugs »

justjosias wrote: post #458
naikrovek wrote: post #320

Earlier quoted context omitted.

I want to know what stuff you guys are putting in public GitHub FOSS repos that you don't want replicated in any way... I also want to know why people think their code is so special that no one else could have ever come up with it independently. Each and every opponent of Copilot is the best developer ever, I guess? That said, I don't understand the choice to use GPL for any reason, so maybe I'm not equipped to under…

Also note: Copilot violates the attribution requirements of permissive licenses like MIT as well. Even if you put your code on GitHub with the intent of it being freely used in proprietary software, attribution is still a fair demand.

That assumes the code snippets being copyrightable and nor fair use.

Re: Give up GitHub: The time has come

#462

Post by smcameron »

naikrovek wrote: post #320
josephcsible wrote: post #122

> If it is, as you claim, permissible to train the model (and allow users to generate code based on that model) on any code whatsoever and not be bound by any licensing terms, why did you choose to only train Copilot's model on FOSS? For example, why are your Microsoft Windows and Office codebases not in your training set? This is my favorite question about Copilot ever.

I want to know what stuff you guys are putting in public GitHub FOSS repos that you don't want replicated in any way... I also want to know why people think their code is so special that no one else could have ever come up with it independently. Each and every opponent of Copilot is the best developer ever, I guess? That said, I don't understand the choice to use GPL for any reason, so maybe I'm not equipped to under…

> I want to know what stuff you guys are putting in public GitHub FOSS repos that you don't want replicated in any way...

What a disingenous reply. FOSS licenses do not grant ability to replicate "in any way" that you wish. You still have to comply with the license terms. What the hell is wrong with you?

> I don't understand the choice to use GPL for any reason ...

The reason is people like YOU.

Re: Give up GitHub: The time has come

#463

Post by pabs3 »

JoshTriplett wrote: post #353
naikrovek wrote: post #320

Earlier quoted context omitted.

I want to know what stuff you guys are putting in public GitHub FOSS repos that you don't want replicated in any way... I also want to know why people think their code is so special that no one else could have ever come up with it independently. Each and every opponent of Copilot is the best developer ever, I guess? That said, I don't understand the choice to use GPL for any reason, so maybe I'm not equipped to under…

> I want to know what stuff you guys are putting in public GitHub FOSS repos that you don't want replicated in any way... Please feel free to use my code in any way that its license permits: attribution for the permissive licenses, share-and-share-alike for the copyleft licenses. Those license terms are the price of the code, no different from a proprietary product's "this costs $x" or "this costs $x/month". I'm happ…

If copyright goes away, that won't magically make the source code of all proprietary software public. The only thing that will be liberated is existing shared-source software.

Re: Give up GitHub: The time has come

#464

Post by pabs3 »

janosdebugs wrote: post #459
natefinch wrote: post #441

Earlier quoted context omitted.

Some interesting reading: https://felixreda.eu/2021/07/github-copilot-is-not-infringin... https://fossa.com/blog/analyzing-legal-implications-github-c...

There's one thing I'm missing from all these discussions and posts: is the generated code even copyrightable? IANAL, but code snippets often fall under the "scènes à faire" doctrine (everybody would do it in a similar way), in which case it's not. https://en.m.wikipedia.org/wiki/Sc%C3%A8nes_%C3%A0_faire

GitHub seems to think it is copyrightable, personally I doubt it is, simply because a human didn't create it and the process it was created by was automatic with no creativity.

Re: Give up GitHub: The time has come

#465

Post by pabs3 »

naikrovek wrote: post #450
skjoldr wrote: post #404

Earlier quoted context omitted.

Then Microsoft should write this as a legal statement on their part that they will take responsibility for. But I doubt they will ever do that.

Microsoft is not the author of the software that copilot helps produce. the person sitting at the keyboard using copilot is the author.

How is copilot not the author?

Re: Give up GitHub: The time has come

#466

Post by bilekas »

naikrovek wrote: post #450
skjoldr wrote: post #404

Earlier quoted context omitted.

Then Microsoft should write this as a legal statement on their part that they will take responsibility for. But I doubt they will ever do that.

Microsoft is not the author of the software that copilot helps produce. the person sitting at the keyboard using copilot is the author.

This is a bit like saying if you hire a freelancer to do some work then you're the author of that work. I'm not too sure i agree with that.

Re: Give up GitHub: The time has come

#467

Post by pabs3 »

zvr wrote: post #349
seadan83 wrote: post #328

Earlier quoted context omitted.

It sounds like Github needs to index the license of where it obtained code, and check the license of a project using co-pilot so that it only suggests GPLv3 code for compatible projects. I'm not sure that it is really enough, enforcement is tough, it could be like one of those "are you 18" checkboxes. I get the impression that this article and it's reaction is largely a lot of people wanting to cry big brother and mo…

I think the article starts from the position "everyone should respect the software license" and obviously promoting the SFC views. You are correct that the problem is hard, because adding licensing info to training data and subsequently use it is something not yet accomplished by anyone (who at least has spoken publicly about it). It might be the only way out would be to have different training sets according to lice…

Amazon claim their Copilot equivalent knows about license info.

Re: Give up GitHub: The time has come

#469

Post by isodev »

smt88 wrote: post #437
galleywest200 wrote: post #433

Earlier quoted context omitted.

I had thought that JetBrains products requires a licence, unlike VSCode which is free in regards to monetary cost. However - the main gripe this community has with VSCode is that it is only partially open source. Microsoft adds extra bits to the code that is not open - mainly the items around telemetry. You can get around this with VSCodium or VSCode-OSS, but those are arguably forks and not a MS product.

I still don't understand why that's an issue. VSCode is a proprietary product. There are tons of FOSS alternatives if it doesn't work for someone or the telemetry is a problem. My criticism would only be that the telemetry is not obvious enough for a casual downloader. I don't use VSCode but I'm sure a lot of people do without knowing about it.

The main selling point of VSCode is their extension ecosystem. Most extensions are open source / created by the community. In essence, Microsoft channels FOSS work by third parties in order to create a pipeline for upselling their products and services.

Re: Give up GitHub: The time has come

#470

Post by leereeves »

Vetch wrote: post #201
josephcsible wrote: post #122

> If it is, as you claim, permissible to train the model (and allow users to generate code based on that model) on any code whatsoever and not be bound by any licensing terms, why did you choose to only train Copilot's model on FOSS? For example, why are your Microsoft Windows and Office codebases not in your training set? This is my favorite question about Copilot ever.

Github's position doesn't appear to offer any advantage with regards to Copilot's creation. OpenAI Codex (which copilot grew out of IIRC), Amazon and Salesforce versions of Copilot exist. Huggingface Bloom was trained on a sizeable amount of public code. Tab9, now behind, was one of the earliest to combine public code repositories with Deep learning for smarter autocomplete. The data requirements for Transformer scal…

Is anyone using Dall-E, Imagen, or any other generative model for art to create commercial products? If so, they're probably also concerned about copyright issues.

CoPilot is being offered for widespread commercial use, so it's held to a higher standard. Respecting copyright is much more important when you're building a business and not just sharing fun AI art on social media.

Post reply on HN