Ken Arnold's Blog

November 2004 Archives

Ludicrous as a Balm for Patent Idiocy

One phrase I am known for repeating is "You're thinking too hard." I use it to mean that someone is looking for a complicated, subtle meaning instead of a simple one. Often it seems that if you pretend the solution is simple, it becomes so.

So I have to say to PJ: "You're thinking too hard." Here are her concerns (quoted from her blog):

1. You'd have to define a clear line in the sand, a definition of obviousness that couldn't be stepped beyond unless it was on purpose, and I don't see how you could. Maybe you do. But no law can be so vague that it's impossible to know precisely when you are breaking it. How could you define clearly enough where that line is?
2. Who do you punish?

This is the level of precision of most legal standards. These words are then interpreted by judges to apply to particular instances, which nails it down harder and harder. But the word "trivial" (or the phrase "trivially obvious") fits into the structure quite nicely.

As for prior art, I would say a patent is "ludicrous" if it steps on prior art "widely known to those normally skilled in the art." In our current poster child, I would say that finding folks who have heard of "not equal" operators for comparing objects, pointers, or references would be pretty damn easy. It's probably hard to find normally skilled folks who haven't.

Liability is even easier. Those who are in a position to profit from the fraud are liable. Typically this is the patent's actual owner, who may or may not be an inventor. In our particular poster child, this would be MicroSoft. I use the phrase "actual owner" because people who are clever should not escape their responsibility by hiding the patent in a shell company or something. Like with any other fraud, you cannot escape liability easily. If you buy a patent, you buy the risk. So don't buy a patent that is likely to be ludicrous.

Others have worried about the new lawsuits involved. First, remember that if you file a claim on a ludicrous patent, you only get paid if you win. This is real incentive to avoid that. And there are protections in the legal system against frivolous suits, which would equally apply here. Remember, the idea is to leverage the existing systems to address the problem, not invent an entirely new system. Standard safeguards and rules apply. We're just defining a new form of fraud, based on existing standards for fraud and patents. Once we've set up the language reasonably, most things should just follow naturally.

I don't declare for or against software patents here. We can leave that for some other time. But politics is the art of the possible, and I think that this suggestion actually is slightly near possible. And I think it would make a difference. And, like PJ, I like the idea of someone getting legally smacked for being "ludicrous". That alone would be worth it.

Self-enforcing Patent Sanity

There are many suggestions for fixing the patent system, from abolishing it to radical surgery. Many of these are praiseworthy in design, but most suffer from a severe problem -- You can't get there from here. Wonderful end results are nearly impossible to attain because the forces defending the status quo are powerful and have little stake in the resulting system.

So here's an idea that actually gives some major players a strong stake in the outcome and could have a big impact: Declare that ludicrously obviously invalid patents are a form of fraud. And enforce that by giving anyone who proves patent fraud by ludicrosity gets paid triple their costs as a reward, plus any damages they can show were caused by the issuance of the patent.

As a starting point, I would define a "ludicrous" patent as one that any practitioner normally skilled in the art would recognize as having prior art. MicroSoft's is our most recent poster child, who seems to be seeking a patent on an IsNot operator that checks if two pointers point to the same place in memory. In other words, it's like C's "p1 != p2" where p1 and p2 are pointers. Uh, hello? Can we say prior art, like almost any machine instruction set? And can we say "obvious" as in "done in almost any language you can name that has pointers or references"?

I model this on "whistle-blower" laws that encourage private citizens to ferret out fraud against the government by giving successful ferreters a chunk of the recovered money. This means that the gov't itself doesn't have to do anything to ensure that a bunch of investigators are always looking for fraud against it. If someone has a good case, the perps pay the investigator. And it also keeps the number of frivolous fraud suits down, because a frivolous fraud suit gets you nothing.

Here we create a herd of patent lawyers who will invest their own money knocking down patents that are ludicrous and collecting bucks. Sometimes they'll get a nice payment from a simple letter that shows the holder that they will lose the case. Sometimes they will win big bucks when someone defends one of them.

It also creates a liability for filing ludicrous patents, which (beyond filing costs) there is none now. It helps protect every major player who isn't using patents as offensive weapons. And another set of major players -- the patent lawyers -- have new ways to make money.

To be fair to those holding patents the predate this rule, I would give holders one year to voluntarily void any patents they hold, but after that, all unexpired patents would be fair game.

So what so you? How about we pay the lawyers to sue the lawyers to keep the pool cleaner?

This work is licensed under a Creative Commons License.