IBM's S/390 Linux guru on the Open Source patent question
An old interview, but a good one...
Following yesterday's piece on Microsoft playing the patent card against Linux we were pointed to an interview with IBM's Karl-Heinz Strassemeyer at Linux Forum Denmark. It's been around for a while, but Strassemeyer has some interesting things to say about IBM, Linux and patents, so it's still worth looking at.
Strassemeyer, from IBM Boeblingen, Germany, is best known for getting Linux running on the IBM S/390, and this is what he used as his first example. He had arranged to see Linus Torvalds in order to show him the S/390 patch, and explain how it worked, "so that he could read into it a little bit and give us advice, on how to do it differently or whatever, right?"
When IBM's lawyers heard about it they said to him this was going to be "a distribution," and that he couldn't do it. This, says Strassemeyer, is how he learned what a distribution was. If however he showed Torvalds the code running on IBM premises, and Torvalds didn't take anything away afterwards, then this wasn't a distribution, and was therefore cool with the lawyers.
This is where Strassemeyer gets interesting with respect to Microsoft's line on Linux patents: "When we put source code into the Open Source community, the code we are putting in there, we have clearly patent clearence for in the sense we know what we put out in the Open Source community so we don't claim own patents on [it] anyway. Our patents clearance process makes sure we are not infringing the patents of anybody else. We are doing it with our proprietary code, so we are fine."
But IBM can't do that with Linux: "What is wrong about this distribution, is basically the millions of lines of code that we never have seen. We don't know if there are any patent infringements [in this code] with somebody we don't know. We don't want to take the risk of being sued for a patent infringement. That is why we don't do distributions, and that's why we have distributors. Because distributors are not so much exposed as we are. So that's the basic deal as I understand it."
Similar reasoning explains his second example of IBM using a proprietary kernel in a particular hardware product, rather than going the vanilla Linux route. "We didn't want to do a distribution, because we didn't want a patent infringement being detected. If somebody would have taken us to court we might have had to stop shipping our product."
Strassemeyer tentatively suggests that some form of 'put up or shut up' patent screening process might provide a solution to this problem, but he concedes he's not a lawyer. And indeed it's difficult to see how this could work with Open Source in general; although there are areas where it's possible to make a shrewd guess about the likelihood of an infringement suit, you can never know what such suits might be directed at.
Which of course goes for all software. But the difference is where we came in - IBM knows all of the software it produced itself, the patents it holds and the patents it has applied for. So it's willing (not exactly willing, it is obliged) to take responsibility. With Open Source software it distributes, however, its lawyers won't let it take responsibility.
A transcript of the Strassemeyer interview is available here, where you'll also find a link to an audio version. ®
Sponsored: Benefits from the lessons learned in HPC