LWN: Comments on "IBM's patent pledge"

show me your ideas!

xoddam — 2005-01-27T05:43:06+00:00

Wrong attitude. Here are 500 patents you can read to the greatest level
of detail and *learn* something from, without fear that you'll be sued if
you use your knowledge.

The patent you *might* be sued for infringing -- you'd be better off not
reading.

IBM's patent pledge

Wol — 2005-01-21T08:46:30+00:00

"In the description of infringing uses, simply state that writing, distributing, and using software doesn't infringe patents if the software is under an OSS license."

STUFF "if the software is under an OSS license."

Just make it "if it's software". Which is the current legal position. And, according to all the *reasons* given for harmonising EU practice, is *not* *supposed* *to* *change*. So let's just make it explicit!

The problem is, the wording of the directive has a clear, already accepted, legal meaning which is very different from the intended meaning as used in the directive. Which is why everybody is saying that the *effect* of the directive will be very different from the *intent* of the directive.

Cheers,
Wol

Re: export regulations

mmarsh — 2005-01-19T01:53:12+00:00

True, but at present cryptographic algorithms (with keys above a certain length) are controlled by the export regulations, and are only "freely distributable" because of specific license exceptions. It would presumably be more difficult for Commerce to decide that something not yet under export controls should be controlled than to change the rules for something already controlled.

The funny thing is that, as good as RSA with keys lengths above 512 bits is, there are much better techniques for some things that aren't controlled at all. There's no restriction on one-time pads, for one thing, and Shamir's secret sharing is as secure as the length of your message space. As in there's nothing to invert, so you can't even brute-force it.

Re: export regulations

roelofs — 2005-01-19T01:34:24+00:00

of course the Commerce Department could change the rules tomorrow. They're the ones who grant the exception, and they have the power to remove it.

In principle, they could do the same for all software, not just crypto code. That is, they technically have the same power to restrict exports much more broadly, and they could "grant" such a restriction at any time. (Of course, politically that would be suicide, but the point is that there aren't really any guarantees for any of us...)

Greg

Re: export regulations

mmarsh — 2005-01-18T20:11:54+00:00

That's correct, at present, though it's "TSU" (EAR 740.13(e) -- be glad that you don't have that memorized). It appears that any FLOSS code would be covered (IANAL), but of course the Commerce Department could change the rules tomorrow. They're the ones who grant the exception, and they have the power to remove it.

Re: export regulations

roelofs — 2005-01-18T19:05:34+00:00

Specifically, if you're writing code in the U.S. and include certain algorithms, your code falls under the Commerce Department's export regulations. If you don't qualify for the "Unrestricted" license exception, you might have to keep track of exactly who is getting copies of your code and ensuring that nobody from certain nations is. I believe this situation would impact any company developing what would otherwise be open source software according to the accepted definition, but I'm not positive.

The only such software with which I'm familiar relates to cryptographic algorithms, and there's a separate exemption ("TSE," I think) for that (in the context of open-source software). See this for details, for example.

Greg

IBM's "open source" definition

gleef — 2005-01-16T06:08:47+00:00

If IBM's license covers all Free Software, Open Source Software, and happens to also include some software that I don't consider Free, I'm still a happy camper.

IBM's patent pledge

error27 — 2005-01-14T21:44:36+00:00

I would definately support that.

Software patents that only hurt^H^H^H^H affected close source software would be great.

To me limitations on what software you can publish infringe on free speech rights and so software patents as currently implemented are unconsitutional. I'd support modifications that exempted free beer and Free speech software publication.

IBM's patent pledge

iabervon — 2005-01-14T21:12:46+00:00

For the purposes of determining whether a particular activity is using Open Source Software, it's relatively simple. You simply ask whether the user or, in the case of a transfer, the recipient gets the FSF's "four freedoms". Since the patent law applies to each use and doesn't have to be carried along the chain of sublicensing and distribution, a lot of the complexity of particular licenses doesn't matter. A lot of the possibilities for licenses which are "Free" for the purposes of users but which let the owner of the original code take modifications proprietary would be protected by this clause, but that wouldn't matter too much if a proprietary version would then lose the protection.

IBM licensed their patent to all software or only a fraction of Free Software

ber — 2005-01-14T18:36:44+00:00

One reason why an patent exception for all Free Software does not work is that if you can fully use patents under a non-protecting Free Software license like MIT/X it can also be used in proprietary applications. To use the patent in proprietary software: Only put the implementation of the algorithm under MIT/X and keep the rest proprietary following the license terms.

So what does mean for IBM? Either they just gave up their 500 patents completely because of the MIT/X trick; or they actually mean pure Free Software solutions, effectively limiting the use to strong licenses like the GNU GPL.

Because it does not make sense to write down terms when you want to completely give up the patents, IBM certainly means only a fraction of Free Software licenses can be used as before when using their patents, any Lesser GNU GPL code cannot be linked into proprietary applications anymore without getting a target for ligitation.

There is even another posion hidden in IBM's patents for Free Software and this is the freedom to learn and use the knowledge in other applications. Okay, that is true for all patents, but usually you do not have the freedom to work with the code and know some might think it is okay.

Conclusion: This might be more than just a smoke screen from the known friend of software patents IBM.

IBM's patent pledge

pm101 — 2005-01-14T16:42:51+00:00

In that case, perhaps we could count on IBM to support a European patent directive that explicitly excluded non-commerical use, free software licenses, acadamic use, and so on? It would need to be carefully worded, so that I cannot just abstract my patented code into an LGPL library, and link it into a proprietary program, but to allow commercial entities to sell free software with patents, but I think it could be done. IBM gets patent protection in the proprietary world, while free software continues unimpinged.

The devil is in the details

filipjoelsson — 2005-01-14T16:41:04+00:00

From the beginning of this article:
"According to the statement, IBM has indicated it will not assert any of the 500 patents against distributors of open source software, so long as the distributing party does not file lawsuits using patents or other intellectual property rights against open source software."

So, if I have a patent - I can ask for money, but if the other ditributor will not give me any - I can't sue. It is not enough that I too license 500 patents to open source, or a proportional part of my patent portfolio (compared to IBM). No, if I want to sue for some reason - I have to ask IBM for permission.

The only way out of this license mess, is if an equally strong player as IBM take them up on the offer - and license another 500 patents. Preferably patents that the licensor knows that IBM is already using - without paying for the priviledge. Now, there are not that many such players - but perhaps Intel, AMD or Novell are strong enough?

IBM's patent pledge

hingo — 2005-01-14T07:38:25+00:00

But then you would need to have a EU directive which defines what is Open Source and what is not. Or, if we use the current practice (which they wouldn't) the OSI would become an EU agency doing this. So in short, it's just not gonna happen.

Missing end-quote

Ross — 2005-01-14T05:20:14+00:00

'Mueller also called it a "diversionary tactic, which may be accurate given
IBM's support of the European Patent Directive that has been denounced by
many of the leading members of the open source community.'

Somewhere in there there should be another '"'. I think it is after the
word tactic.

IBM's right to revoke

giraffedata — 2005-01-14T01:48:48+00:00

Actually, IBM's pledge says IBM's right to revoke is based on whether the patent user files a lawsuit, not asserts rights.

And I'm sorry to see that. It reinforces the mistaken impression many people have that lawsuits are bad things and inequities are caused by the availability of a court system.

Lawsuits are good. They protect little guys from big guys; nice people from immoral ones. The problems that get ascribed to lawsuits are actually due to the underlying liability that allows one to win a lawsuit. All a lawsuit is is a way to get money from someone who owes you (at least in your opinion) and won't pay.

IBM's pledge creates a bizarre situation where if Company A writes a letter to Company B saying, "Hey, you're using our patent. Pay up" then to exploit IBM's generosity, Company B has to say, "Screw you. I'm stealing your invention and you can't stop me!" I.e. he has to make Company A sue him even though he knows Company A is right. He has to be immoral. It would be better if IBM said "asserts rights" instead of "files lawsuit," because then the letter would never come in the first place. In fact, upon request Company A would probably have to send a letter saying, "you're welcome to use our patent."

The devil is in the details

brouhaha — 2005-01-14T00:15:26+00:00

Because if the second company tries to assert one of its own patents against an open source project, IBM will withdraw that company's right to use the IBM patents.

IBM's "open source" definition

mmarsh — 2005-01-13T18:20:42+00:00

For some software, this is about as open as it can get. Specifically, if you're writing code in the U.S. and include certain algorithms, your code falls under the Commerce Department's export regulations. If you don't qualify for the "Unrestricted" license exception, you might have to keep track of exactly who is getting copies of your code and ensuring that nobody from certain nations is. I believe this situation would impact any company developing what would otherwise be open source software according to the accepted definition, but I'm not positive. Fortunately, I've only ever had to worry about software arising from academic research.

The devil is in the details

rgoates — 2005-01-13T17:53:05+00:00

I don't understand your argument. How does a company's use of one of the 500 IBM patents in an open source project effectively license all that company's own patents for open source?

IBM's patent pledge

iabervon — 2005-01-13T17:11:06+00:00

Personally, I think this is a PR move to counter the FUD about OSS being (more) vulnerable to patents. They want to get the press to have headlines about a whole bunch of patents being not a problem for OSS, and maybe text in articles about IBM suggesting that it might crush anyone who tries to cause problems for OSS.

I wonder if IBM is going to try to get an exemption for OSS into a next EU software patent law. In the description of infringing uses, simply state that writing, distributing, and using software doesn't infringe patents if the software is under an OSS license. I wouldn't be too surprised if they were to try this, because OSS is generally unproductive to enforce patents against (even where this would be legal) and OSS is the main opposition to software patents.

The devil is in the details

ncm — 2005-01-13T15:18:58+00:00

This is something Lucent tried when they released Plan 9 under similar conditions. Arguably, they doomed Plan 9 by it. By the time Lucent released it under a really-free license, the hardware base had moved on, Plan 9 had already been dismissed from most potential adopters' minds, and Linux had matured.

IBM's game is different enough that it might turn out better for them, but this sort of trick tends to bite back in surprising ways.

RCU missing

ncm — 2005-01-13T14:39:22+00:00

That RCU is not among the relaxed patents is telling. It must have been discussed at length. It appears they decided they don't want to risk having the BSDs' NUMA performance approach Linux, even if it meant they could sell more chips to Apple. With a bit of work, Apple could run Macosix on top of Linux instead of Darwin, or (with a lot less work) just run Linux on its server installations. Maybe IBM would like that batter. It shows that IBM still considers the OS market an arena for competition, and not wholly a commons.

The devil is in the details

filipjoelsson — 2005-01-13T11:31:09+00:00

So, a high profile company with patents of their own decide to take IBM up on the offer - and include some IBM patented technology in an open source project. (Perhaps the engineer doing so thinks; "hey, IBM licensed this for open source projects - I don't have to ask the legal dpmnt for this tiny detail.")

By doing so, the company has effectively licensed _all_ of their patent portfolio to open source projects. IOW IBM withholds all but 500 patents, but in doing so they deny anyone who take them up on their offer the ability to withhold anything at all.

This is a very unbalanced license.

let's assume sincere intentions for a while...

pointwood — 2005-01-13T10:30:07+00:00

Well, software patents are probably the biggest threat to open source software and the open source community should therefore fight it at all cost.

IBM still supports software patents. I'm not that conserned about IBM suing any open source developers, but IBM is not the only company with software patents. There's a certain company located in Redmond that isn't quite as big an open source fan as IBM is.

It might be well-meant gesture from IBM, but it is of little use in reality.

let's assume sincere intentions for a while...

nettings — 2005-01-13T07:45:01+00:00

while i agree with the author that the outcome of ibm's move remains to be seen, i think we should consider it a well-meant gesture for now.

my guess is that inside ibm, there are open-source-minded folk and followers of the traditional ideas of intellectual property, and we should not view ibm as a monolithic entity with just one opinion.

the opening of just 500 patents sure is an ambiguous move given how big blue continues to be the most active purveyor of software patents in the industry.

however, it might be a great achievement for those inside ibm who have grasped the concepts of freedom in programming and are trying to steer company politics towards open-source-friendliness.

i'm all for healthy scepticism, but let's not put our allies at ibm off by being too vitriolic about their first steps...

IBM's "open source" definition

torsten — 2005-01-13T05:35:38+00:00

Who cares about the 500 they won't sue for, show me the one they are going to sue for.

IBM's "open source" definition

ncm — 2005-01-13T04:17:35+00:00

IBM's definition of "open source" here is a bit odd. For example, it includes licenses that require the copyright owner to be notified before each redistribution, and the recipients identified. That definition might suffice for the purposes of this patent grant, but it would be an unfortunate if it were repeated in other contexts.