Personal FAQ

Software Patents

Too Many Patents for Too Little Innovation

I've heard a lot of people cite the RSA as something that deserves a patent. I agree and often cite it myself. But there are so few that do...

I think most reasonable people would have voted RSA as a winner to this award, so it passes that litmus test for a proper system. I think most reasonable people would not vote for the XOR patent to get a multi-year protection, even though it was cool in its time--the cost of development was just not steep enough, so it passes that litmus test as well. And no matter how strongly people on the committee felt that something intermittently important like LZW compression was, it would still be available for others to use freely in 5 years, not 17, which is a worst case most of us could live with.

And, no matter what you think about people "deserving" the patent protection, the most obvious truth is this: because the present system treats reinvention as an infringement (rather than as a proof of obviousness), all programmers are required to know about all patents (to avoid infringing them, since there is no other way to tell you've infringed than to do a patent search on every idea you have). This is just impossible from a practical standpoint. Disagreement on this point, at least, must necessarily be seen as ridiculous and/or disingenuous.

Proposal: Make Patents a Rare and Prestigious Award, like the Nobel Prize

Why not do something innovative and just elect a blue ribbon panel of judges to give away patents based on how cool they are and how useful they are? It would be like the "Nobel prize for Innovation".

The award could have two components:

• A right of unique use for a period of time (not more than 5 years)

  The period of time would be set by the judges, and surely shorter than 17 years, but probably a function of the development costs, so that more expensive developments could get somewhat longer protections. The time periods should not to exceed 5 years. but some might get only 6 months or a year.

  These shortened time periods reflect the rapid obsolescence of technology and the need of the community to continually build on each others works.

  This component of the award is an incentive to the business or funding agency that paid for the development, since it means that there is a degree of retroactive subsidy of the work for the winners.

• A monetary award to the inventor(s).

  This component of the award need not be proportional to the cost of development. It is merely symbolic in nature, but would encourage businesses to name the individuals who did the work. A business should not be eligible for a patent award at all (including the right of unique use component) if it did not designate which individuals were responsible.

The Benefits of Finite Numbers of Awards

If few enough awards were given (e.g., ten or twenty at most), and if their duration was five years or fewer, rational people could keep track of what was patented and what not. Expensive patent searches would no longer be necessary.

It might seem that a very small number of awards would leave out companies. But, in fact, empirical observation shows that most software engineers (at least the ones I have known over my career) don't believe most software patents are justified because the bar for "obviousness" is way too low and in most cases people believe that any good engineer working on the same issue would eventually reach the same conclusion, so offering a patent as an incentive to get there is wholly unnecessary.

If few awards were given, there would be fewer nuisance lawsuits over trivial and obvious matters, since the trivial and obvious matters would stand no chance of winning.

If few awards were given, the practical likelihood of being sure you were going to get one would be low enough that you would not bank your business on receiving one. As such, the need to offer 17 years to recover an investment would be unnecessary, since no one would be likely to invest so much on the promise of patent alone.

Does this mean an end to invention? No, because copyright and trade secret already provide adequate means of protecting real investment. If an idea is so obvious that as soon as you see it, it can be copied without copying the code, it isn't much of an invention. If it would take a while to work out the details, it is naturally protected by copyright and trade secret. So, I allege, this change would mostly change the structure of investment, it would not cause an end to investment.