According to Red Hat's new policy on software patents, "In the interests of our company and in an attempt to protect and promote the open source community, Red Hat has elected to adopt this same stance," which is to adopt defensive patents. "We do so reluctantly because of the perceived inconsistency with our stance against software patents; however, prudence dictates this position," Red Hat stresses.
I'd like to reassure everyone. Red Hat's patent claim is very weak. iMatix first proposed content-based exchanges in the AMQP draft specs two years before Red Hat filed this. Further, the actual design is poor: dynamic routing XML is obvious but slow. There are faster ways to do this - extract the essential routing data from the XML and turn into a topic key or headers.
But the quality of the patent is actually irrelevant. What we have seen is that there is a risk that participants in the AMQP process will silently file patents on it. which would be detrimental to all of us, users and open source vendors alike. Even if this particular patent is contested and rejected, there will be others.
"Red Hat's patent policy says 'In an attempt to protect and promote the open source community, Red Hat has elected to... develop a corresponding portfolio of software patents for defensive purposes. We do so reluctantly...' Meanwhile, USPTO Application #: 20090063418, 'Method and an apparatus to deliver messages between applications,' claims a patent on routing messages using an XQuery match, which is an extension of the 'unencumbered' AMQP protocol that Red Hat is helping to make. Is this a defensive patent, or is Red Hat cynically staking out a software patent claim to an obvious extension of AMQP? Is Red Hat's promise to 'refrain from enforcing the infringed patent' against open source a reliable contract, or a trap for the unwary? Given the Microsoft-Red Hat deal in February, are we seeing Red Hat's 'Novell Moment?'" Reader Defeat_Globalism contributes a related story about an international research team who conducted experiments to "quantify the ways patent systems and market forces might influence someone to invent and solve intellectual problems." Their conclusion was that a system which doesn't restrict prizes to the winner provides more motivation for innovation.
Of course, I understand why Red Hat would feel this necessary. But there are other ways around this that are less “’90s Microsoftian.” The most applicable idea is “prior art”. Basically what this means is any information (in any form) made public that is dated and relevant to the patents’ claim of originality can be used to dispute a patent. With that in mind I would much rather see Red Hat (in true open source form) create a sort of “patent wiki” that would post all information relevant to any ideas or technologies they are working on. With dated (and well documented) information, this would serve as strong defense against anyone applying to patent something Red Hat was already working on.
"Small Software companies cannot afford to go to court or pay damages. Who is this software patent system for?" —Marco Schulze, Nightlabs Gmbh
Comments
neo
2009-03-17 13:49:44
Pieter Hintjens
2009-03-17 14:36:51
The news here is that Red Hat sought a patent on an extension to an open standard - AMQP - that it's participating in. It was the way this patent was filed, and the matter it covers, that incensed other members of the AMQP working group.
I'll note that at the time Red Hat filed their patent for an XML routing 'exchange' - August 2007 - they had not yet written software. They were reading the spec and extrapolating into areas that were not covered by AMQP's patent policy but which would be interesting to patent.
Red Hat has not really explained why they did this. Their rep on the working group has said it was to "test the broken US patent system" and to "defend against a known troll" but neither of these explanations make sense.
Further, they told no-one about it. I'd expect that kind of behaviour from some of the working group members (MSFT) but certainly not from Red Hat.
Why is patenting around an open standard such a crime? Because it is a standard way to ambush the market. You spot an area that is not covered by the IP policy, you patent it, and when customers realize they need to extend into that area, you quash all competition and demand your licenses.
Again, standard operating procedure for many firms. Not for Red Hat.
What makes the situation more spicy is that Red Hat helped draft the AMQP IP policies. So they knew the loopholes.
Some people have said that if Red Hat invented this, they have the right to patent it. To answer that: the AMQP spec which Red Hat read in 2007 was largely my invention: I dreamed up exchanges and bindings, hammered them into shape with my team and the guys at JPMorganChase, explained how they should work, explained how to extend AMQP with custom exchanges, wrote thousands of pages of design notes, RFCs, and diagrams that finally condensed - by my hand and over three years - into the AMQP spec that Red Hat read in 2007.
Yet it's Red Hat that claims a patent on a trivial and obvious extension to the spec, in an area where it is clear that people will need to develop.
That is the news: a stupid, hypocritical move by a firm that has a lot more to lose than others by filing software patents around an open standard.
People have also said that Red Hat will never sue open source projects. This is not the point. We've seen some outright lies from Red Hat about AMQP - including one press release where they claimed to have invented it - and we know that they are desperate to sell MRG, their AMQP package. Patents are mostly used for that: FUD, to scare potential clients away from the competition. It is very rare to see real litigation.
People have said this is a defensive patent. Well, firstly it cannot be: it is a patent on specific functionality. The only possible infringers are Red Hat's direct competitors. RabbitMQ, OpenAMQ, ZeroMQ...? Secondly, the notion of a "defensive patent" is marketing. Perhaps a "patent to trade with another large firm" would be more honest. And such practices are not ethical. Patent deals are very close to cartels. Those who refuse to take out patents, or are too small to afford the inevitable litigation they cause, are excluded and turned into clients.
And this is the point: Red Hat, instead of competing on quality, seem to be taking the same patent FUD route of other firms. "Competitor product X infringes on 237 of our patents. We're not going to tell you which one." That would be a tragedy and the community needs to remind Red Hat that software patents are not an option when it comes to selling software.
Sorry for the long comment. Thanks for reporting on this.
Pieter Hintjens
2009-03-17 14:42:09
There is also a project called Peer2Patent which is looking at new ways to bring prior art to the attention of examiners.
However, your point that it is hard to fight patents by pulling up prior art is accurate: the patent system is much more friendly to its clients than those who would get in the way.
Roy Schestowitz
2009-03-17 16:44:19
Thanks for the detailed explanation.
Ars Technica has some coverage now too.
Roy Schestowitz
2009-03-17 17:48:56
Roy Schestowitz
2009-03-17 19:18:25
I've only just seen it and this happens to agree with my interpretation that Slashdot sensationalised.
neo
2009-03-17 20:03:37
"here is a simple and cheap way to file prior art at the USPTO called a Statutory Invention Registration, which shows up on examiners’ searches and prevents the risk of patent trolls."
You should be well aware that patents filed by Red Hat are not just to prevent others from claiming rights but also counteract potential patent threats. Prior art doesn't serve that purpose.
Actual patent trolls have no easy defense at all since they don't sell any products are not worried about getting counter sued. It is a expensive process and that's why patent system is broken
Roy Schestowitz
2009-03-17 20:13:32
"Defensive" patents don't help against patent trolls. Period. That's why OIN is moot in many cases.
ram
2009-03-18 00:08:32
My company did some defensive patents. What you do is patent the technology, and make the disclosure very extensive. When the patent is granted the disclosure, references, and of course claims, all appears in the USPO database. On the first renewal date one can let the patent expire, but now the information in it is firmly in the public domain and 'prior art' that is hard to ignore.
neo
2009-03-18 01:42:12
You are merely repeating what I said. However OIN is still useful. There are lot of companies filing patent suits while not being patent trolls. You need to update your badly written article and correct the information.
Pieter Hintjens
2009-03-18 08:27:05
To comment on PJ's remark... RH's promise to not sue FOSS applications is irrelevant to an open standard. Any open standard needs adoption by commercial closed source applications as well.
Secondly, people keep mentioning OIN as if the only threat here was Microsoft and IV. OIN is a *Linux* patent pool. It is for defending Linux against very specific threats. Nothing about OIN will protect an open standard.
And lastly, this is about an open standard. Think about that for a second. An open standard, and a firm taking patents on essential and obvious extensions to that standard. Unscrupulous at best. Being inside the AMQP process, I can say that Red Hat did this in secret and it looks very clearly like an attempt to own the space. To come back, as they did in the press release, with non-answers, suggests it is a deliberate move.
Note again: Red Hat's patent, on an extension of AMQP, is not covered by the AMQP agreements, and clearly prevents any closed source firm from implementing XML routing with AMQP.
That is not about fighting off patent trolls. Neither is it about self-defence from attack. It is about using patents to block competition. Further, it is a first strike, thus aggressive.
Irrespective of a firm's past behaviour and stated intentions, it is what happens on the ground that counts. I have all respect for Red Hat but their behaviour here is unarguably wrong.
Secret patents on open standards are unethical. They defend no-one. They damage the standard. They scare off adopters. They provoke an arms race. We've seen this a hundred times.
I'm really surprised PJ is not treating this more seriously.
Pieter Hintjens
2009-03-18 08:36:20
"You should be well aware that patents filed by Red Hat are not just to prevent others from claiming rights but also counteract potential patent threats. Prior art doesn’t serve that purpose."
This is what Red Hat tell us. But examine this argument. Any patent an exclusive right to some "invention", i.e. some space in the market. Red Hat's patent claims exclusive rights to do XML routing over AMQP. They have promised to share that space with FOSS developers. (A cynic would say: that's because Red Hat make their money by repackaging FOSS code).
Now you say this can counteract a potential patent threat. Well, another patent might try to occupy the same space, or a different space. If it tries to occupy the same space, registered prior art is the fastest, cheapest, and most reliable antidote.
If the threat patent occupies a different space, then claiming this XML-over-AMQP space has zero effect. Zero. The two patents exist independently.
Right? You understand that the claim that taking a patent will prevent a threat from other patents is bogus, and misleading?
So what, then, is this patent good for? Only two things. (a) preventing real competitors from entering that space. Namely, Microsoft, Novell, IBM, and the many closed-source firms who today do XML routing and would love to put AMQP into the mix. (b) trading with other patent holders, so that Red Hat can gain access to some other space that is currently closed off to them.
Neither of these two scenarios helps AMQP, neither helps FOSS implementors of AMQP, and neither helps the FOSS community. Who this patent does help is Red Hat, which is why they took it.
Once again, please examine the facts on the ground and put aside the marketing and rhetoric that companies issue. Sadly Red Hat answers to its shareholders, not the FOSS community. It makes dollars, and does exactly what's needed to keep its free labour force happy, but not a jot more.
Pieter Hintjens
2009-03-18 08:50:53
It's a common belief that software patents would be fine if they only affected closed source products. But all software, closed or open, contributes. When we develop new standards, we need the closed implementations as much as we need the open ones. Even if I only make, and use, open source, I'll defend the right of closed source firms to compete, free of software patents.
Red Hat seem to see software patents as a fair weapon in a fight with their competitors. They sponsor "peer to patent", which is an attempt to make better, cheaper patents, not eliminate patents. In Europe, Red Hat have stopped working with the FFII to end software patents. They seem happy with the status quo.
So. Facts on the ground. A software firm secretly patents around open standards. Claims patents so it can do deals with other patent holders. Prefers to license patents rather than fight them. Does not fight software patents where it actually could (in Europe). Treats patents as a way of discriminating between FOSS and closed source. Invests in "improving" the patent system by making it cheaper to get more unbreakable software patents.
Am I speaking about Red Hat or Microsoft? It is hard to tell any longer, except that Red Hat are happy for me to write their code for them.
We'll see how the story with AMQP develops. If Red Hat have filed more patents around it, we know what stripes the tiger is wearing.
[1] http://www.press.redhat.com/2009/03/17/discouraging-software-patent-lawsuits/
Roy Schestowitz
2009-03-18 09:28:57
Pieter Hintjens
2009-03-18 09:59:45
Short answer: no.
A longer explanation...
The background to this is that the pro-patent lobby has been trying to get software patents legalized in Europe for a decade, first by modifying the European Patent Convention, then by the 2005 Software Patent Directive, then by EPLA[1] in 2006, and now through the Community Patent and by unilateral decision in the EPO. Each time it's the patent industry, EPO, lobbyists and certain software firms (mainly MSFT) claiming that Europe's small IT firms will die unless they can patent all their work. (The irony is rich but not sweet.)
On the other side, the abolitionists, primarily the FFII and friends in the small IT sector, with little money and just endless sacrifice. Pretty much every person who's worked in the FFII has ended up bankrupt, burnt-out, and wrecked. When I was president I tried to raise funds to pay the core activists a living wage. That worked but only for a while. My own business got hammered as I spent two full years on FFII work.
Now, Mark Webbink, the former Chief Counsel of Red Hat helped us with a number of initiatives including most vitally the European Patent Conference [2], which was a series of major events that brought together abolitionists from all sectors, and some very high up. Mark presented the abolitionist case at many conferences. Red Hat joined in many FFII campaigns and workgroups.
In mid-2007 Mark left Red Hat to join the SFLC and in 2008 Rob Tiller took over. From that point, all cooperation with the FFII stopped and my attempts to restart it failed. In a number of key areas, such as a review of the Symbian DLL patent in the UK, Red Hat decided to work alone, ignoring the community.
I can state for the record that Red Hat have not donated a single Euro to the FFII in 2008 or 2009. Yet this is the volunteer organization that was and is most significant in stopping software patents in Europe, the largest economy in the world.
Thus, their claims to be against the software patent system need to be measured against their actual acts. Words are cheap. Deeds and dollars count.
Perhaps this also explains why I am so utterly, profoundly, disgusted by Red Hat's patent claims on AMQP, a protocol that represents several years of my life.
[1] http://epla.ffii.org [2] http://www.eupaco.org
Roy Schestowitz
2009-03-18 11:04:50
Pieter Hintjens
2009-03-18 11:25:24
I might. But I prefer to add my comments to your blog. :-)
Matt is being generous to Red Hat, which is fair enough. He's not demonstrated a great understanding of the facts, which is normal: the case is complex and subtle.
In any case the truth will emerge. Either this was a singular mistake by an over-enthusiastic lawyer in Red Hat, and they are fumbling their response. Or it is part of a deliberate move to own AMQP, and there will be more such patents in the pipeline. If it's a mistake, it'll all go away. If it's deliberate, all hell will break loose when the next patent pops, and I'm sure Matt will be at the front asking "what the heck is this?"
Like I said, patents around open standards are a special kind of nasty.
Roy Schestowitz
2009-03-18 11:39:21