[jolmg]

I think the GPL is a good license, both v2 and v3, for the restrictions they place to promote more FOSS and ensure software users the rights I would hope everyone believes they ought to have (e.g. the 4 freedoms to the software of devices they own via the anti-tivoization clauses).

Having said that, do most licensors use the GPL as opposed to licenses like BSD/MIT? And of those that use the GPL, do they do it for the restrictions it has as opposed to just following a collective habit?

Looking at what I have installed on the computer I'm on, GPL is hanging in there. I see:

  $ pacman -Qq | xargs pacman -Qi | grep -Po 'Licenses *: \K.*' | sed -E 's/ +/\n/g' | sed 's/-.*//' | sort | uniq -c | sort -n | awk '$1 > 100'
      242 MIT
      277 LGPL
      348 GPL
      381 BSD

However, looking at https://github.blog/open-source/open-source-license-usage-on..., I see:

  |  1 | MIT          | 44.69% |
  |  2 | Other        | 15.68% |
  |  3 | GPLv2        | 12.96% |
  |  4 | Apache       | 11.19% |
  |  5 | GPLv3        |  8.88% |
  |  6 | BSD 3-clause |  4.53% |
  |  7 | Unlicense    |  1.87% |
  |  8 | BSD 2-clause |  1.70% |
  |  9 | LGPLv3       |  1.30% |
  | 10 | AGPLv3       |  1.05% |

[badlibrarian]

The broader point being that every one of those licenses is just that -- a license. The terms of the license apply because the material is copyrighted.

And it's one place where you can directly specify your intent. In your license, say that everything reverts to the public domain in 5 years or 10 years. Grep away and show me how many licenses do that.

Varying durations for different types of media should be discussed as part of copyright reform. But simple statements like "10 years" reveal that people haven't thought things through.

[wat10000]

Or maybe they have thought things through and they just don’t agree with your conclusion.

For people who choose a non-viral license, why not go straight to public domain? I see three reasons: 1. it avoids confusion and difficulty with countries that don’t recognize public domain. 2. it provides an explicit disclaimer of liability. 3. people like the requirement to credit the author or distributing organization.

1 wouldn’t be a problem with short copyright terms. 2 shouldn’t be either. I doubt someone would get anywhere trying to sue for damages caused by a defect in copyright-expired code. You’d lose 3 after 10 years but I’d guess open source authors see that as a nice-to-have rather than a hard requirement. The credit in proprietary software using non-viral open source is almost always buried in some “licenses” file nobody ever looks at anyway.

[jolmg]

> In your license, say that everything reverts to the public domain in 5 years or 10 years. Grep away and show me how many licenses do that.

Look, I'm no lawyer, but my broader point is that something like that might not make much if any difference to most. It doesn't seem to me that there's much difference between the MIT license and public domain. The MIT just requires attribution and propagation of the license text.

If you add up the MIT licensed projects with others that have similar licenses, you might get to a 51%, at least according to the GitHub stats. I would think most of these people just picked a license by what other people picked. They don't really, really care to put the particular restrictions they did.

I'm not saying that 10 years is a good number, or that licenses are bad. I'm just saying that your pick of FOSS might be a poor example to argue about the need for long copyright terms.

The only ones among the FOSS community that likely care to have long copyright terms are those that pick GPL-type licenses, which have more substantial restrictions to ensure the freedoms of end-users.

[badlibrarian]

Linux, Blender, and WordPress immediately spring to mind as software that would be in a very different place if their codebases reverted to public domain at the 10th year of their existence.

[jolmg]

In what particular way do you think they'd be different if they'd gone into the public domain at 10 years?

[shiroiushi]

The Linux kernel has changed a lot in the last 10 years. Having all the code in it that's >= 10yo become public domain would only mean you'd be able to run an ancient kernel on old hardware without worrying about the GPL license terms.

[bluGill]

How many are still running kermels from 10 years ago. even for mainline stuff with insignificant changes such that it is out of copyright (an interesting legal question itself), there is enough that is significant in new kernels

[badlibrarian]

It's not about running a 10 year old Kernel, it's about a trillion dollar corporation owning a source snapshot, throwing 5,000 engineers at it, and not contributing anything back.

It also effectively turns GPL3 to GPL2 on a rolling 10 year basis.

People freaked about Tivo 20 years ago. Now imagine what kind of chaos Nvidia and Oracle could cause starting from even Ubuntu 14 or a 3.18 Kernel.

[bluGill]

The continued success of tye various bsd proves your fears are way over stated.