Hacker News Reader

Comment by Mayzie | Hacker News Reader

So, Facebook is the only company that struggles with this and come up with this arcane solution?

Google and Microsoft are just as big, yet they can manage open source perfectly fine without this ridiculous licence.

So speaking as the guy responsible for Google's opensource-licensing for a long time (i've finally been able to divest myself of most of it :P):

You are right in a sense, but Google/MS also take more risk to do so.

Google did also try something not quite like this, but in the same vein.

The initial webm license terminated both patents and copyright if you sued over infringement. (Apache, et al just terminate patent licenses. You could still use the software if it had no patents).

The practical upshot you couldn't use that software if you sued over infringement.

This was done, of course, because of the state of the video codec world.

Fortunately/unfortunately (depending on who you ask) despite being okay with it prior to release (IE we asked, ran it by the a lot of folks, etc), certain partners later decided they weren't okay with it because it wasn't GPLv2 (but was GPLv3) compatible.

So we changed it.

The point of all this is that: License innovation by itself is not a bad thing. Needs change of communities change over time, including open source. IMHO, innovation in this space needs to be done for a good reason, and needs to be done collaboratively, if you want any chance at success.

If you can answer, I have one question. My understanding is that Angular uses an MIT license, without any patent grant. I think this means that, if Google holds a patent that covers some mechanism used in Angular, and my company uses Angular, technically Google could sue me. Is this correct?

> they can manage open source perfectly fine

It's easy to "manage open source perfectly fine" if you just open-source less of your code, or release it without any patent grant at all. I'm not sure how much credit should be given for those non-efforts, though.

No explicit patent grant has been functionally sufficient for decades so not sure what you're saying there.

What about Google and Microsoft's FOSS contributions do you consider "non-efforts"?

> What about Google and Microsoft's FOSS contributions do you consider "non-efforts"?

It's not the FOSS contributions that are non-efforts. It's the things that were never released at all. What percentage of engineers at Microsoft work on code that has been or ever will be open source? How is releasing 10% of your code with no patent grant better than releasing 20% of your code with a conditional one?

Also, "no explicit patent grant has been sufficient" is no excuse for bashing a grant that has exactly that as its worst case. Any non-zero protection is still greater than zero. By way of analogy, would you yell at people on your team who you felt weren't working hard enough, then give a big pat on the back to one who played video games the whole time? Well, maybe you would, but an objective observer might think it seemed unfair. They might even wonder about what other motivations might be involved.

So, until all (or some large quantity) of software produced by an organization is open source, any software produced by an organization that is open source should not be considered worthwhile efforts?

Not what I said at all. I'm just saying they shouldn't get more credit for a non-release than for a release. Apparently the HN attitude is:

* Release lots of software with a conditional patent grant: awful.

* Release lots of software with no grant (e.g. BSD): OK.

* Release very little software, and that under the no-grant license (e.g. MS): you rock.

Sorry, but that's completely f-ing backward. It's literally impossible to get to there from open-source principles, so what what principles are people starting from?

Though untested, many think the BSD and similar licenses provide an implicit grant that can't be strategically revoked. Future case law to the contrary would make a lot of open software unusable or at least a legal minefield. Providing an explicit grant with explicit conditions of termination makes it easy for Facebook to selectively restrict use of anything that might be under patent in ANY of their projects. At best it's not in the spirit of FOSS and at worst it asymmetrically weaponizes open licenses.

It's true that, practically, most companies won't ever have to worry about Facebook infringing their patents, but I think it's garbage on principle to reserve first-strike litigation while making it impossible for someone else to do the same and use software you've made available under a supposedly permissive license.

> Though untested

That phrase alone should swing the "legal risk" argument the other way that how it's being portrayed here.

I've never worked at Facebook (so I can't compare directly), but I've never met a Googler who shied away from open sourcing code for legal reasons; no one ever said "I would open source this, but only if I could attach a patent grant to it."

There are plenty of reasons that someone might choose not to open source code (e.g. they like working on only internal infrastructure or don't want to bother with external maintenance), but I don't think the patent grant has anything to do with why a developer at {Google, Facebook, Microsoft} would choose to open source (or not) a library.

I've never worked at Google either, but I have worked at other companies where people got in serious trouble for trying to open-source something that was associated with the company's IP. It's important to remember here that copyrights and patents are separate. A company might be very permissive when it comes to copyright (though woe betide the engineer who tries to open-source something related to trade secret IP) while still aggressively protecting their patents. Have you seen a Googler actually try to attach an explicit patent license to Google code? After Oracle, I'd be pretty surprised. The BSD/MIT non-license is the real stealth option here. It precludes no action by the patent holder, under any conditions, so of course your local legal department would have no problem with it.

Google defaults to Apache.

The whole process is documented here (https://opensource.google.com/docs/releasing/approval/) but here's the tl;dr:

1. Googler tells their director that they want to open source some code; the director approves it.

2. Googler tells the Open Source + Patents Office that they're releasing code. The OSPO says to use Apache as the license unless you have a good reason not to, and does a sanity check to ensure that the license information is in all the right places in the repo.

3. Googler posts the repo to GitHub.

Googlers don't usually even think about licenses. The OSPO says to use Apache, so that's what we use.

Do you suppose that neither the director nor the OSPO is checking which patents relate to the code under consideration, and that neither might reject the request on the basis that it would compromise valuable IP? It would actually be irresponsible for them not to. Given the value of that IP to the company, they have a duty to ensure that it's protected. The fact (if fact it is) that individual engineers don't worry about it means approximately nothing except that they're sheltered little pups.

Reply on news.ycombinator.com