Microsoft Can’t Undo Bilski Damage

Microsoft was at the Supreme Court yesterday, in the ironic position of trying to invalidate a patent.

It's the i4i patent on XML, and while the Microsoft claim is based on prior art, the issue before the court involves the standard of proof it must meet in proving a patent is invalid after it is granted.

In fact, this is a case about the speed of law vs. the speed of technology.

The patent in question actually dates from 1994, as the Web was being spun. The idea of "custom XML" originated with a French SGML editor in 1991. The patent itself was granted in 1998.

What's really at issue is that i4i bought the French inventor (left) for its patent while Microsoft hired its XML 1.0 co-editor from the same French outfit, known as Grif. The custom XML concept was embedded in Microsoft Office starting with Office 98.

All this happened when Microsoft was still trying to avoid software patents through legal agreements saying that, if it were doing business with you, you wouldn't assert patent rights against it. Unfortunately it couldn't sign such a deal with Grif because Grif was gone by this time and Microsoft had hired the Grif guy. It wasn't until this century that Microsoft hired an IBM patent attorney (later author of Burning the Ships) who decided Microsoft should grab for all the patents it could, then use them for cross-licensing its way into other technologies (like Linux).

Got it? Microsoft hated software patents when this technology was first created. It got a creator instead.

In this case, i4i acted as a standard patent troll. They sued in Texas' "rocket docket" (the eastern district in Tyler) in 2007, won their case in 2009, and it's before the Supremes two years later. Microsoft's case is that it cobbled together new evidence for a re-examination of the patent, and while i4i won in that re-examination the "burden of proof" used by the Patent Office should have been lowered from "clear and convincing evidence" to a  "preponderance of the evidence" standard.

That question, the standard by which granted patents should be invalidated, is what was before the court yesterday.

Kremlinologists who listened to the oral arguments heard liberal justices like Stephen Breyer making arguments that supported the patent holder (and its higher standard of proof), with conservatives like Samuel Alito making arguments supporting Microsoft (with its lower standard). The Chief Justice, John Roberts, is recusing himself from this case, so it could (in theory) wind up a 4-4 tie.

If I were to guess, it would be that Justice Elena Kagan will be given the task of drafting a narrow decision that could invalidate this patent without changing the standard (as in Bilski vs. Kappos, where the patent was tossed but the court refused to change rules for the future).

The court doesn't want to change patent law, but at the same time it has to recognize that the law in this case is incredibly far behind the technology. It should also recognize that courts are to blame for this, because they created the idea of patenting math — software is a set of mathematical algorithms — in the first place.

Think of it. A Canadian troll wants to stop sales of Microsoft Office, going back to 1998, over a technology originally created for SGML back in 1991 that's now essential glue for the Web.

We can't have that. But the Supremes don't think fixing the problem is their job, it's that of the Congress. And Congress is too divided to do that job.

Yet another reason why courts should not have ruled software patents valid in the first place.