Open Sourcers Drop Software Religion for Common Sense | Hacker News Reader

Open Sourcers Drop Software Religion for Common Sense

wired.com

19 pointsrevorad31 comments

Ironically, Sleepycat couldn't have executed without the GPL; it's a fundamental part of their business model, which relies on dual licensing.

There's more to the GPL than "religion". When you release code under BSD, MIT, or Apache, you give up all control of it. Anyone is free to commercialize it without adding any value. The better marketing & sales team can and will beat the origin of the software in the market.

If you're not sure whether you might someday want to make money from your code, GPL is the more pragmatic choice. You can always relicense BSD later --- but BSD, once picked, is forever.

That. GPL is excellent for a "take it and contribute back, or pay my next holiday - either are fine by me!". It's a very fair deal.

I'm not saying everybody should go GPL, I'm just saying it's great for many purposes. I came the other way from the open source zealots. I used to scream "viral effect!" whenever something was GPL, thought it was bad, bad, bad. Nowadays, when something is dual licensed GPL and commercial, I tend to become a happy customer.

Typically, the long-term prospects are lot better than with BSD software slammed on github with a "here you go" note attached. No evidence, btw, just a feeling.

The way this has come about with relation to web and internet software is that the actual applications being built are closed source but a lot of the libraries and commodity pieces of software are being open sourced with permissive licenses.

for eg. Twitter, Google, Facebook etc. produces and consumer a lot of open source software but their core products are still close source. You can segment out portions of your software stack and release them for free while still retaining a competitive advantage with your core application. This isn't an all-or-nothing decision and this hybrid approach is the most common.

I think Twitter is a good example. They both built and scaled up using open source and also contributed their own open source software, which has been improved upon by others and in-turn benefited Twitter. It didn't require a license where contribution was essential and nor did it require giving up any competitive advantage (i'd argue it solidified their position).

I'm not saying there's anything wrong with MIT licensing code, but again: MIT licensing concedes more from the author to the public, and is an irrevocable grant, so from a purely pragmatic standpoint, "if you have to ask, GPL; at least that way you can change your mind later".

Also, hopefully it goes without saying: what's good for a ridiculously fast-growing startup surfing on a tsunami of VC money is not necessarily good for a solo practitioner.

This is an unfortunate pan of GPL and a clearly business-sided discussion of licensing. Developers use GPL to keep their work in the open and encourages everyone working on a project to share their work. On the other hand, "permissive" licenses are well received by businesses because the work produced under BSD-like licenses are perceived as "free labor".

So here's the rub. Businesses want free labor, but they want to be paid by their customers. Is that capitalism or socialism?

It's Capitalism if the labor isn't forced. Capitalism allows for citizens to willingly participate in any sort of arrangement. Sharing is just another form or voluntary participation in the system.

Socialism uses the monopolistic threat of violence and incarceration to force arrangements predetermined by the State.

See the difference?

This was fairly comprehensively debunked in a talk at FOSDEM 2012:

http://fosdem.org/2012/schedule/event/is_copyleft_being_fram...

http://linux.slashdot.org/story/12/03/03/142229/gpl-copyleft...

Black Duck Software don't publish their methodology, whereas John Sullivan does publish his (based on studying the license of Debian packages).

The "practicality" argument doesn't hold water with me: none of it matters without freedom. If you care about code, the GPL gives you freedom by requiring that people give back in the manner they take: with code. It is only necessary because people weren't giving back.

I can understand that someone might want to shirk the quid-pro-quo responsibilities of the GPL, but I won't be paying for that kind of anti-social product.

"freedom by requiring" < see the contradiction? true freedom has no requirements or demands.

The free software movement aims to protect the freedom of the users, not developers (although the user could be a developer). Developers are in a position to control their users unfairly, so providing them the freedom under a permissive license does nothing but to further endanger the freedom of the user.

This is an important distinction. Those who focus on the freedom of other developers miss the point of the free software movement. Consider an individual looking to rob a bank with a pistol (as opposed to robbing a user's freedom with their own software). Before the robbery, they notice a permissive amory across the street. Your permissively licensed software is that armory. You are empowering that individual to blow a hole through the bank rather than hold someone at gunpoint.

Your software, as great as it is, can also be used to lure in users for exploitation. Nothing like a heavily armed robber with candy!

I do, but I care about CODE. Without code, I didn't GET freedom. :)

True freedom also doesn't exist. The mere act of publishing something carries with it an implicit claim of copyright. Thus, the GPL substitutes one set of restrictions for another.

It's the old "you can catch more flies with honey rather than vinegar".

The GPL is the strongest vinegar to businesses that may not mind participating in open source, but don't want to have their business assets dependent upon something so restrictive.

As a developer you have a trade-off decision to make: Choose the license that restricts the use of your creation in hopes that it forces others to share, or choose a license that may end up being used for private commercial reasons - but will likely be more widely adopted because it's less restrictive.

The beauty of the whole system in my mind is that each developer is free to make his own choice.

Nothing wrong with choice. I think it's important that people treat me fairly. That's practicality, IMHO.

I guess. That's kind of the decision that a lot of console or heavy copy protection game producers make: We want to get paid for every copy of our game that is played.

Then a lot of no copy protection PC game producers make a different decision: We want as many people as possible to play our game. Some will pirate it and not pay us. We hope that many more will do the right thing and pay for it if they like it.

Yep. Luckily the sky is rising! :) http://www.techdirt.com/articles/20120129/17272817580/sky-is...

This is why I use the MPL.

even more pronounced when you look at web software. look at the most watched projects on Github and play spot the GPL license:

https://github.com/popular/watched

I couldn't count a single project. I have never used the GPL in any software I write because I want anybody to be able to use my software and contributions.

Google, Microsoft, Apple (who are actively purging what remains of GPL code) et al will not adapt your project for bundling or distribution or use if it is GPL. I think most developers are more interested in having users of their software than attempting to shoehorn them into some ideology, which is reflected in the license usage stats you see.

to add to that, the FSF completely screwed up v3 which likely scared away anybody who was on the fence about which way to go (it was also more to do with the anti-commercial rhetoric around the changes as it had to do with the changes themselves)

I know there isn't an easy alternative to it, but that Github list is pretty wonky; things like impress.js and Devise are at the top of it, but are clearly not among the most important pieces of open source software. Similarly, I review other people's Rails apps for a living and have never seen a project use that particular top-of-the-list form builder.

I can only conclude that the number of Github users "watching" a particular repository is not a particularly useful metric for how good/important/popular that software is.

Finally, for obvious reasons, Github has a pronounced Ruby/Rails bias, and that community already has settled norms on licensing.

I noticed that - the fork list isn't a good representation either. It would be interesting to get a sample across github, bitbucket, google code and whatever that site is called that .NET developer use.

I don't think a survey of all of those site s combined would fall far from the 90-95% BSD/MIT license rate you see in the GitHub sample (again, talking about web related open source software).

Anecdotally, I have also noticed that many Android/iOS/mobile libraries are also MIT/Apache/BSD.

Like another commenter noted, the FSF (wow what a bad logo) recently did a survey of Debian packages that rebuts the 90-95% claim.

But there's no way to apples-apples this anyways. Communities like Rails produce starbursts of itsy bitsy little packages like form-building gems, and have MIT licensing norms. You'd need to normalize somehow for value or code base size.

What I'd say is that "GPL use is plummeting" is an extraordinary claim, that extraordinary evidence has not been marshaled to support it, and that a more likely phenomenon is that the GPL is roughly as strong as it's ever been in its original domain (the licensing of large, complex, expensive- to- produce software packages) and probably not as prevalent in some of the newer more... uh... casual... software development domains.

It's funny you point to GitHub considering git is GPLv2 licensed itself. I imagine many of the tools which make those projects possible are also GPL licensed.

OC should have said "GPLv3" instead of "GPL". Most companies are pretty comfortable with GPLv2.

jQuery Diaspora Linux

jQuery is also MIT licensed, which is why Microsoft was able to bundle it.

large parts of Diaspora are also dual-licensed (not sure what the breakdown is).

I am still going through the list and haven't found a single GPL-only licensed project

What about Linux? That said, it might be the exception that proves the rule!

how about Git itself?

focused on internet and webapp software and the trend there, which is overwhelmingly tilted towards the permissive licenses. much more than the 50/50 split suggested in the OP.

so the Q is why is the GPL getting so little adoption amongst web developers.

I think the article is using wrong examples if they talk about web companies such as Facebook, Google or Twitter.

The GPL doesn't force you to free code you don't distribute, for example the code behind your server. This means that, if Facebook did use a GPLed library to, say, show you advertisemnts based on your 'likes', they wouldn't have to GPL their modifications (if any) or the part of proprietary software that used that specific library, let alone their whole infrastructure.

So this argument simply doesn't apply to this kind of software.

A license that forces you to release your modifications even if you don't distribute your software (again, a web application for example) is the Afero GPL, that is just another beast.

TL;DR If you don't distribute or sell software, the GPL is not a concern.

It's strange to see that now, inasmuch as years ago, when I worked at VA Linux Systems, the general attitude was of philosophical affinity with Open Source pragmatism. It wasn't dogmatic religion. I still think that's sensible and it always seemed to me like many also felt this way...and the GPL was not necessarily a hindrance to that. It was simply one tool among several in the licensing toolbox.

The article mentions the drop since June 2008... AKA, a year after the GPLv3 came out.

But sadly, not a mention of the controversy of the new version was made (directly). Some interesting points about the change in focus were made, but I think that v3 played a large part.

Reply on news.ycombinator.com