I’ve been doing some research on the issues in Microsoft’s lawsuit against Tom-Tom. Here’s what I’ve found about the patents are at issue in the case:
|Car Navigation Systems
|Jan 16 2001
|Vehicle Computer System with Open Platform Architecture
|May 30 2006
|Method and System For Generating Driving Directions
|Oct 3 2006
|Portable Computing Device-integrated Appliance
|Mar 13 2001
|Vehicle Computer System with Wireless Internet Connectivity
|User Interface Model
|Mar 9 2004
|Methods and Arrangements for Interacting with Controllable Objects within a Graphical User Interface Environment Using Various Input Mechanisms
|Nov 26 1996
|Nov 28 2006
|Common Name Space for Long and Short Filenames
|May 26 1998
|Oct 10 2006
|Common Name Space for Long and Short Filenames
|Flash Memory Management
|Jul 30 2001
|Method and System for File System Management Using a Flash-Erasable, Programmable, Read-only Memory
First, some discussion of scope…
As you read through these, the first thing that will become apparent is that the first four, the car navigation system patents, are no threat to Linux whatsoever. They’re very specifically about certain particular combinations of hardware and software features. I’m not going to discuss them much further except to note with amusement the amount of bare-faced gall anyone from Microsoft would have to muster to pretend that their technology is “open platform”.
The content of the UI-model patent (‘032, as Microsoft’s lawyers nicknamed it in the brief) is a little more troubling. Press reports have tended to lump it in with the car-navigation patents, but it could be read to apply to lots of other types of systems; note for example the references to keyboards and joysticks. The language is dense and vague, but I read it as attempting to cover any situation is used to control the way that (other) hardware input devices are connected to pieces of software and/or configurable hardware in the system. I don’t see any aim at Linux here, but I think the game-console manufacturers should consider this one a threat.
Most of the public attention has focused on the two FAT patents. Interestingly, these are not patents on FAT itself. Rather, they have to do with methods for translating between long filenames and the DOS-style 8+3 names that FAT still uses internally. They’ll read on any implementation of FAT that wants to present long names to the user, including open-source ones.
The flash-memory one could be the biggest worry in the bunch. It seems to be claiming things that any flash file system needs to do to manage its hardware. No threat to Linux on its own hardware, but it might be deployable, if upheld, to block anyone from shipping in the U.S. a Linux filesystem that manages flash devices, whether it’s FAT-compatible or not.
I’m sorry to say that I don’t think any of these patents can be struck on grounds of obviousness. There might be an outside chance of successfully attacking 032 that way, but the PTO’s application of the “obviousness” test is notoriously likely to be fooled by claims language that sounds more complex and arcane than the techniques it actually describes. Besides, two of the important ones have been rexamined, pre-Bilski; if the PTO were going to find obviousness in these at all it would likely have happened then.
Previously undiscovered prior art could come out of nowhere and scupper any of these, of course, but my best best judgment as an engineer in related fields is that this attack is most likely to be effective where it isn’t very helpful, in the car-navigation patents. I see a slightly better chance of attacking 032 from this direction; configuration of specialized input devices through settings in a GUI is something X programs do all the time. The problem would be convincing a court that the similarity is strong enough.
UPDATE: Harald Weite claims that the FAT patents have been invalidated in some jurisdictions by prior art. I knew they had been struck down in Germany, which is doubtless the jurisdiction Weite speaks of; he claims the prior art in question was the Rock Ridge extensions to the ISO9660 standard, supporting long names on CD-ROM filesystems. This is plausible.
In the present situation, the interesting question to ask is whether these would pass the Bilski filter. For those of you who have better things to do than follow developments in IP law, “In re Bilski” is a recent decision by United States Court of Appeals for the Federal Circuit in a lawsuit challenging a business-method patent. The Court invalidated the patent, ruling that for an idea to be patentable, it must â€œ(1) it is tied to a particular machine or apparatus, or (2) it transforms a particular article into a different state or thing.â€ (direct quote from the decision).
The judges in Bilski declined to rule on the patentability of software (the case was over a business-method patent, and judges generally prefer to rule narrowly rather than broadly lest they be reversed) but they did cite a decision indicating that simply running a program on a general-purpose computer is not a patentable process. These is now a vigorous debate going on over what those two tests actually mean for software parents.
It seems to me that 032 and 642 are highly likely to fail whatever version of the Bilski test emerges from that debate; the claims are pretty abstract, and it’s hard to see how they can be considered tied to a machine or a transformation of material objects without an interpretation of those requirements that would render Bilski meaningless. (Microsoft, of course, has already argued for such an interpretation in an amicus brief on that case.)
By contrast, the car-navigation patents seem the most likely to survive in re Bilski precisely because they’re the most closely tied to a specific hardware context. But the open-source community has no reason (other than generalized hostility to software patents) that it should care.
The FAT patents fall in between. They certainly fail the transformation test — the only things transformed are the (non-material) names of files in a filesystem. Whether they’re defensible therefore depends on what courts construe the language about being “tied to a machine” to mean. The stakes are high, because any version of the Bilski test that neuters these patents would probably render most software other than device firmware unpatentable.
UPDATE: Good analysis by Steve Vaughn-Nichols here.