The wires are full of news around Linux and Patents, with OSRM claiming that Linux infringes lots of ’em, and IBM promising not to litigate ’em. Well, I go way back on this issue; herewith a software patent war story, flavored with the usual cynicism.
50-Gig War Story · A decade or so ago I was working for Open Text Corporation, a company I co-founded, and at that time we were mostly in the search-software business. We were 32-bit software but I’d figured out a hack to search billions of words by running parallel.
One day I got a call from the CEO saying “Tim, you gotta go visit Microsoft right now. We can maybe sell them software for a fifty-gig database installation.” I laughed “Tom, get real, nobody has fifty gigabytes of text.” (This was before the Web). He said “They bought the full text of all the U.S. Patents.” Well, OK then.
So the Microsoft guys built this mind-boggling system the size of a dozen fridges with 50 or so rack-mounts from Compaq or Dell or someone (Blades? Who knew from blades?), running the then-brand-new NT4.
Getting it all going was quite a story, but it’s not the story I’m telling here. That story started late on the evening when, alone in a lab in Redmond, for the first time I had an index of the whole thing, and I sat down to try it out. Since search software is what I know about, I used the search software to search for patents about search software. An hour later, I was sick and stunned and horrified. Every basic component of conventional search software, including pieces dating back to the Seventies, was patented. Postings list? Patented. Boolean queries? Patented. Tokenization (patented) of Asian languages (patented) using finite automata (patented) with embedded markup? All patented. Search with parallel servers (gasp)? Patented. Mostly by people I’d never heard of, but IBM and Xerox and Intel and lots of other big names were there.
I can’t imagine that search is that much different from any other area of software. Thus the lesson.
The Lesson · In software, assume that everything is already patented. You can’t build anything, no matter how new it is, without infringing someone’s patent.
Of course, a very high proportion of those patents could quite likely be invalidated by prior art, from both the patent space and outside. But that’s expensive; a patent is an extremely potent offensive weapon in the hands of a smart attorney, because most companies don’t have enough financial slack to fight intellectual-property litigation.
GNU/Linux · Yes, of course it infringes hundreds of patents. Every significant piece of software does. Everybody who has experience in the software-patent space, even less extreme than mine, has known that for years. Every competent VC in the world knows that. So the current hustle and bustle is interesting, but there’s no new news.
OSRM · So, who is Open Source Risk Management anyhow? Their web site is remarkably uninformative about their corporate structure and ownership; it’s nice to see that Bruce Perens is on board (PDF), so let’s give them the benefit of the doubt. But I gotta say, the recent publicity blast (PDF) about “Linux potentially infringes 283 patents” is relatively news-free and can’t possibly be accurate. Why? Because when people say “Linux” they mean this huge sprawling agglomeration of software; yes, the kernel, but libraries stacked on libraries stacked on libraries stacked on compilers and tools, building the stairway from dumb silicon to GUIs and applications. I totally guarantee that there are way more than 283 patents that could put a gleam in an an aggressive attorney’s eyes.
IBM · It’s nice for an IBM guy to get up on a stage and say they won’t litigate in the kernel space; but I think of all that software sprawl and I think of IBM’s current multibillion-dollar patent-licensing revenue stream, and I think of the stress around competing with other companies waving the Linux flag, and I don’t think the issue has gone away. Heh, I see Jonathan is on the case too.
The Real Problem · Maybe I’m naive, but I don’t really worry that much about IBM or Microsoft litigating Linux into the ground. First, the wave of outrage they’d provoke, and the PR damage they’d take, would likely be intolerable for a public company.
Second, the Linux community would—after some pain—figure out a way to route around the litigation; it would be real work, but it would happen. So litigation wouldn’t be a good long-term strategy for a big player.
I worry more about intellectual-property specialists looking to monetize dormant patent portfolios (the world has lots more companies like SCO); in particular, one of them going after some part of the Linux sprawl that’s simultaneously user-visible and has a high lines-of-code-to-developer ratio; I have some particular pieces in mind.
But what I really worry about is the basic broken-ness of the US Patent system. It’s sucking money and time and resources out of the whole technology ecosystem; it’s a bleeding sore on our business and our culture and our people. It’s way past time to fix it.