Hacker News Reader

Comment by PoignardAzur | Hacker News Reader

I'm very frustrated by some of the answers here and on Twitter. Carmack says:

> The GPL virality wound up being a net negative, and more value would have come from BSD. [...] The best aspects of GPL work didn't manifest, but tons of opportunities to just copy-paste-modify were lost due to license concerns.

One of the answers on Twitter:

> Then what if it's being exploited by big techs and make your work proprietary? GPL preserves the four essential freedoms, BSD does not.

Yes. Okay. Carmack is aware of that.

When he says "The best aspects of GPL work didn't manifest", he specifically addressing those concerns, and saying that he basically saw no benefit from "not being exploited by big techs".

Like, I find it intensely frustrating that, when an industry veteran says "I've heard the pros and cons of each approach and I found that in my real-life experience GPL was a net negative", people answer with a 101 on the philosophy of GPL.

Is there a word for that? Dismissing empirical evidence by clinging to philosophical assertions?

View on HN

Eh, people are allowed to disagree or challenge Cormack, it doesn't bother me that they brought up that worry despite knowing he is an experienced respected person.

I do find it notable how many of my peer computer programmers/software engineers no longer believe in open source though. 15 years ago it was very popular to believe in open source, as a way of combating the power of giant corporations.

The dislike/distrust/desire to combat large corporations is still there, but there seems to be a popular and growing opinion that, as they have seen big corporations get only more powerful while using open source, this must mean that open source serves big corporations, and we should be anti-open source if we want to combat big corporations. So it's still anti-corporate, but now much more popular to be anti-open source.

The difference I guess is that now the idea is that open source privileges big corporations against smaller for-profit concerns, that it hurts the kind of "entrepreneurs" that the speakers want to be, hurts their ability to become rich themselves. Whereas before the idea was that open source would help "the people" against the corporations, not help some entrepeneurs get rich themselves.

ANYHOW. This is definitely a thing. Anyone who's been in this "scene" for 20 years is seeing it.

Is anyone writing more about it and what's going on?

Carmack is clearly still in favor of open source, he just wishes he had gone with BSD. And it doesn't sound like it's because he's anti-corporate. It sounds like it's because he works for a mega tech corp and his mega tech corp hates the GPL so he has to do a ton of paperwork to use any GPL code there. (which is the same as working for any tech corp.)

Well it’s also because we didn’t see much done with the GPL’d source releases for his engines, just small improvements and standalone mods (The Dark Mod, Xonotic, etc)

No commercial game developer is willing to ship a game with GPL’d code because the license is incompatible with most middleware and console SDKs. We missed out on what could have been a free and open competitor to Unity or Unreal.

The crux of his argument seems to be that he wants to be able to copy and paste other people's code and create derivative work without the GPL carrying along.

Right, I wasn't speaking of Carmack as part of the trend. Carmack still has an old-school relationship to support of open source. I was speaking about the increasing number of people opposed to open source, saying things like that it's clear that the reason someone supports open source is because they "work for a mega tech corp", you illustrated the trend I am speaking about, precisely. Support for open source is now seen as the side of "mega tech corps", and in fact open source is decreasingly popular in online developer communities. With people accusing those of supporting it of doing so only because they have been bought off or tricked by large corporations. Exactly. This is a change.

Focus changed. The 90's was the rise of the PC, the 00's the rise of the web, and the teens rise of the money making mobile platform.

The rise of the PC is what sparked open source as we needed cheaper (free) tools. Once the tools were free it accelerated the growth of the web which was built using said free tools (remember LAMP stacks?). Mobile built on the web and used the same free tools allowing more money to be made. It basically put a computer in everyone's pocket in two decades which is no small feat.

In the end open source did win in a way. We still use open source tools. Linux powers most of the web as well as Android. Much of iOS/MacOS is built of/with free tools (GCC/clang, BSD, etc). But OSS remains a beast of burden which powers money making platforms. And there's a lot of money to be made on those platforms so focus has greatly shifted.

Reminds me of a phrase used to describe the evolution of Jazz music: "Born in a whiskey jar, grew up on pot, and died on heroin." Open source was born in that whiskey jar of general purpose beige boxes. It grew up during the wonky web years and showed great promise. Now its highly addicted to corporate money and slowly fading away.

There was a lot of money to be made on the platforms that OSS ran on from the very beginning of OSS. That's not a change. It wouldn't have been an issue if there wasn't.

In the eyes of original generation of OSS supporters, the 'lot of money to be made' was what made OSS important, to ensure that people access to software that served them as users, instead of just being made to make a lot of money off of them.

Arguably, despite the takeover of open source, that vision failed.

So, now, rather than opposing "money to be made" with the "interest of users" those opposed to open source (while making a case that they are the ones really more "in the original spirit of open source" than open source itself)... say they are sticking up for the ability of the small entrepeneur to make money not just the big corporation.

That's the change, not that there's a lot of money to be made from software. From the start of the OSS movement there was a lot of money to be made off of software, if there weren't OSS wouldn't have been relevant in the first place. Original OSS said if making sure users have the freedom to do what they need with software reduces the money to be made off of it... that's fine. New line says the thing is to reserve the right of the small entrepeneuers (or those who can convince the internet mob that they are smaller and more deserving than Amazon anyway) to make money off users, and try to prevent big corporations from getting a piece of the pie that rightfully belongs to "deserving" businesses, who are deserving because... they're smaller.

I have some ideas of what the differences actually are, I think it's not about how much money is to be made, but how the actual individual software engineers made money; who their employers were and how those employers behaved. I think it has to do with both greater "austerity" and "efficiency" among large employers, and greater dreams of getting rich as an "entrepeneur" among individual developers.

> The best aspects of GPL work didn't manifest

Did someone get sued for downloading a copy of a game with the quake source code in it?

Did someone release a game with quake code in it, only to go after users for patent infringement?

DRM locked devices built on the quake code?

A license is simply the condition for which an author allows others to do things for which copyright law make illegal. There is also a culture and a movement that has been built around common rules for software development.

Carmack does not explain what he wanted out of the license, and so people attempt to fill in the gaps. That people interpret that as "Dismissing empirical evidence" is on those making that interpretation. I do have a word for that, but we should strive to improve the tone on HN and so its better to just explain the purpose of those who replied to Carmack.

(I don't flag comments often but this is a prime example. Hugely frustrated, dismissive, with a bit snarky tone. The comment would be hugely improved by simply skipping the last paragraph. The outcome? A bunch of up and down votes in all the comments below in typical flame war fashion)

>A license is simply the condition [...]

I don't think reducing it to that actually helps the discussion.

A license/law/policy can't be analyzed via its bare verbatim text. One also has to look at its unintended effects in the real world. Yes, the GPL license has noble intentions but that doesn't mean it's immune from causing actions that indirectly contradict its goal. (Wikipedia article on that phenomenon: https://en.wikipedia.org/wiki/Perverse_incentive)

A license or law can have a stated positive goal or ideal with the text written for that intention and motivation ... but the complexity of the real world may work also against it.

>Carmack does not explain what he wanted out of the license,

To use framework of "perverse incentive" to interpret his tweets..

He wrote: "The GPL virality wound up being a net negative," and followed up with:

>the best aspects of GPL work didn't manifest, but tons of opportunities to just copy-paste-modify were lost due to license concerns. It is possible that some of the source ports wouldn't have been as open, but I'm pretty sure there would have been more total users of the code, likely making the amount shared in the open still greater. I'm still supportive of lots of GPL work, but I don't think the restrictions helped in this particular case.

So in this particular case, he claims that GPL unintentionally caused less openness instead of more. Seems like a textbook example of Perverse Incentive.

Yes, we could get pedantic and say GPL verbatim text mentions "freedom" instead of "openness". True, but many interpret GPL's copyleft reciprocal nature of "I share my code and you share your changes back" as trying to create source code abundance. Carmack is claiming that in his case, GPL's restrictions unintentionally caused less code to be shared.

One can still disagree with Carmack but it's still worth understanding what he's claiming.

Looking at the quoted text one could guess his goal is maximized amount of users, alternative maximized number of copy-paste-modify events, and that he thought that the GPL would produce those?

(We could dig into the terminology around "openess", a classic tar trap in the GPL and BSD flame wars, but I doubt such path would lead towards comments getting more thoughtful and substantive.)

Going thus back towards getting users and copy-paste-modify events, as a strategy GPL is rarely if ever a maximize for those goals. The discussions around licenses for stackoverflow code seems to have concluded that claiming copyright at all on small code snippets is counter productive and something most users don't seem to even acknowledge or know that it has a license. A patent grant is however a good protective measure, which would be more akin to the apache license rather than BSD.

For maximum users we have both Iphone and Android as examples of BSD vs GPL, and the conclusion seems to be that the license has very little effect. In contrast one could make a strong correlative argument that proprietary software on top and DRM is the best strategy out there if you want to maximize users.

I doubt however either is what Carmack is talking about. Instead it seems to be the rather common pattern we have seen in the past when old developers test out open source. They look at projects like linux and expect to get similar contributors by releasing their own projects under GPL, and then they end up not getting many and announcing that "the best aspects of GPL work didn't manifest". The question is if that would also have happened under BSD. He could try it today, and if we got the same result, what would that prove?

> One can still disagree with Carmack but it's still worth understanding what he's claiming.

Exactly, one can still disagree, and he is making a claim. Those are 2 things:

He does not know what would have happened, had they gone with BSD license. No one does ultimately. It might be a case of "the grass is always greener on the other side".

Second is, that he might write however good code he wants, but what is its use, if it can only be used in 1 product, and then gets copied into other people's proprietary products, with no modifications shared back to the community? His code could have been "trapped" in that one BSD licensed product, if they did not make use of it themselves later on. By going GPL, at least they got the people, who did in reality, not just in imagination of an alternative past future, use the code, to contribute back. Those are the facts and they were caused by the license choice. A BSD license does not ensure any such thing. Of course it is possible, that there would have been more adopters and more people sharing code back, but that is not guaranteed, which is kind of the point of GPL.

> Of course it is possible, that there would have been more adopters and more people sharing code back, but that is not guaranteed, which is kind of the point of GPL.

BSD/MIT/permissive and our use of them did make it a hell of a lot easier to approach my previous employer about contributing back on company time/pay (they now do). They even came back with "but no GPL." One of the biggest problems in OSS is paid time/contributions, and GPL discouraged it in at least this one case.

Which is a bit weird, considering, that those things, that are contributed back, are not ensured to remain available, when anyone else makes modifications and has the option to make their changes a proprietary thing.

One would think, that if any contributions flowing back are allowed by a company, then it is to GPL licensed projects, because competitors, who want to use it in their products, would have to also contribute back their modifications. Basically to ensure, that a competitor cannot gain an invisible advantage based on ones code. (If all play by the rules.)

One problem with arguing that something is an unintended consequence is that you have to prove that they're not among the intended consequences for the creator. I'm really not convinced that the original GPL authors weren't aware that their license would somewhat reduced openness (relative to permissive licenses) in its pursuit of other goals.

> (I don't flag comments often but this is a prime example. Hugely frustrated, dismissive, with a bit snarky tone. The comment would be hugely improved by simply skipping the last paragraph. The outcome? A bunch of up and down votes in all the comments below in typical flame war fashion)

Fair enough, I guess.

I do think my last line had merit. I'm pointing a pattern I'm seeing: people dismissing Carmack's object-level argument without arguing on the object level themselves, because they deem that the general ideology of GPL makes Carmack's object-level claims irrelevant.

> “Dismissing empirical evidence by clinging to philosophical assertions?”

Dogmatism usually means this.

Ideology fits pretty well, as the term is used in modern times.

In case of Linux, GPL is regarded as a good fit.

In case of games, it'd be surprising now, to see GPL as a good fit. Open source and GPL are not used predominantly, thus never disrupted much in that space. However, this is based on a kind of hindsight bias as well.

In many ways, Free Software is regarded a means by itself: To keep software Free. So the goals of the ideology might not ressonate with the goals of every game developer.

Right. Carmack's goals are clearly not aligned with those of the FSF. The FSF is perfectly happy for Free software to be of no value to those who develop non-Free Software, indeed they actively prefer it that way, viewing all non-Free software as unethical.

Evidently Carmack would rather the Id source releases be of use to as many developers as possible. As he says though, My partners would never have gone for it, precisely because it would have been valuable to developers of non-Free games, who compete with Id.

And to be fair, he even concluded with "I'm still supportive of lots of GPL work, but I don't think the restrictions helped in this particular case."

Carmack explicitly mentions the BSD/GPL question in connection with copilot. He is working on AI now and has an explicit interest in the intellectual property discussions that flared up after copilot.

It's natural that people think he's using his status as a super programmer to lobby for the position that currently benefits him.

I don't see why he would be more qualified to assess the pros/cons of the GPL than the Linux and GCC authors.

> I don't see why he would be more qualified to assess the pros/cons of the GPL than the Linux and GCC authors.

That is simple: he has thought about it in more depth than most of them. Most Linux and GCC authors don't really care about licensing much, just code they can work collaboratively on. Since the code they are working on is GPL they use GPL and get back to work.

I suspect if Linus took the time to really think he would use BSD for Linux - it hasn't been possible to make this type of change since like 1992 at the latest though so there is no point in him thinking about it. And in turn that means my claim is not testable.

In https://youtu.be/PaKIZ7gJlRU Linus talks about BSD licenses at 8:30 (well, somewhere around there).

Its a good watch for the philosophical background on which license Linux is under and Linus's feelings about the GPL v2 and v3.

No, he wouldn't have used the BSD license - it didn't match what he wanted for linux.

> I don't see why he would be more qualified to assess the pros/cons of the GPL than the Linux and GCC authors.

I don't think he is. I would be equally frustrated if Linus Torvalds made an argument about the pros and cons of each license and people replied with generic proselytism.

My point isn't that Carmack's status means people should automatically listen to him. I'm saying he made a thought-out argument, based on practical considerations, and the way most people answered didn't really address the argument, so much as re-state a previous held belief that license X is better than license Y.

> Like, I find it intensely frustrating that, when an industry veteran says "I've heard the pros and cons of each approach and I found that in my real-life experience GPL was a net negative", people answer with a 101 on the philosophy of GPL.

You think that's frustrating, try watching an expert talk about economics.

> You think that's frustrating, try watching an expert talk about economics.

Physics, too. I once mentioned on a forum that at least as the fat guy in my group of friends I coasted down a slope faster on my bicycle than did my skinny friends. (It was quite noticeable, too).

The number of people who jumped in to tell me I was wrong and attempted to do that math using their first semester freshman physics knowledge to prove it was astounding.

They did a decent job--for bicyclists on the Moon. For bicyclists on Earth--not so good. They all neglected wind resistance.

Wind resistance should slow down the larger person more.

Ignoring wind resistance is the problem. Undoubtedly mass was accounted for, but they got the incorrect answer because wind resistance has a much greater effect on the lighter rider. The extra momentum of the heavier rider easily overcomes the small increase in wind resistance. Anyone who has biked with someone of significantly different body mass will have seen ample evidence that bigger people coast downhill faster.

I don't think momentum should affect how fast you slide down. The reasons are different.

I suppose I should have said "inertia". The heavier rider has greater inertia, meaning the wind has less effect on their velocity.

Imagine a spherical cyclist ;-) Mass scales like the cube of its radius. Surface area (air resistance) scales like the square of the radius. As the radius (mass) increases, the air resistance becomes relatively less important.

The force from wind resistance should be more on the larger person. It should be proportional, at a given wind speed, to the ratio of their cross sectional areas.

The acceleration from a given force is inversely proportional to the mass of the body it is applied to.

The larger person has more mass and a larger cross sectional area. The question then is this:

Is Mf/Ms > Af/As,

where Mf and Ms are the masses of the fat and skinny persons, and Af and As are the cross sectional areas of the fat and skinny persons?

If Mf/Ms > Af/As then the fat person will feel less force from wind resistance at a given speed than will the skinny person.

Let's say my mass was 50% higher than my friend's mass. Unless my cross sectional area is also 50% higher than theirs, I'll experience less acceleration from wind resistance than they do at a given velocity.

If my increased mass came from just being bigger in all dimensions, I'd be 14% bigger in all dimensions, and have 31% more cross sectional area.

In reality, the increase in cross sectional area is even less than that. When you get fat you don't get taller. You get wider, but not as much as you get longer. Height and width are the ones that matter for cross sectional area assuming you aren't ridding side saddle.

Given Veritasiums most recent video, the degree doesn't matter, even physics professors can be wrong. The degree is just there for you to be wrong less, you will still be wrong, just less.

Linus Torvalds has similar conversations with gpl advocates regularly. “no, i am happy for companies to use linux this way”. And the people just can’t handle this idea. Like suggesting that god doesn’t exist to someone devout.

It is probably more about a v2 vs v3 debate in the case of Linux. He really likes copyleft GPLv2-style and really thinks it has been a net-positive if not one of the key things for the success of the project.

But Linux is GPL! And Linus himself believes Linux wouldn't have become what it is now if it weren't for GPL.

In the context of Linus values, open source means basically the right of forking software, but copyleft means the right of merging back worthwhile patches from downstream forks. This is absolutely critical in an industry full of binary blobs: this is basically the reason we have any kind of source code for Android kernel drivers, routers, and many other products.

What Linus doesn't value is the goals of free software movement: empowering the user of the computer. He just wants the code. That's why he doesn't care about tivoization, but he cares a lot about GPL.

This was part of the v3 fight - he wanted software free, but also so you were free to use it. Didn't mind that you could use it with locked down hardware (TiVo / cars / phones / med devices / security etc)- still had to release source

> Dismissing empirical evidence by clinging to philosophical assertions?

Small nitpick -- Carmack's assertions are anecdotal, not empirical.

You can very well reason about licenses in a logical (mathematical) manner, so it is a bit strange to speak in terms of "empirical" or "anecdotal".

Maybe you can but the tweet is just Carmack expressing his opinion.

Aren't anecdotes pieces of empirical evidence?

Calling anything and everything “just anecdotes” is HN’s favorite way of dismissing your arguments.