Patent Invention Machines

kryzx writes: "Here's one to tickle your imagination: using genetic programming to come up with new, patentable solutions to problems. Could be happening very soon. Here's an article at MIT Technology Review. This work, being done at Stanford and Genetic Programming Inc. by John Koza and company has already succeeded at reproducing quite a few ideas for existing patents, ranging from old to very recent. It's apparently much easier to compare against existing patents than sift through hundreds of surviving algorithms to determine if they are useful, original, and patentable.) Also, this company is a good target for your tech envy, with their 1,000-node Beowulf-style cluster of Pentium II 350's and 70-node cluster of 533 MHz DEC Alpha's. (There are pix, too. PII cluster on the main page, Alphas here.) Wanna play with the toys? They have job openings for programmers. :-)"

Comments

Work there? No thanks.

[Apuleius]

I'd rather keep my soul.

Re:By Definition, should not be Patentable.

[Anonymous Coward]

Exactly. Excuse me while I get a patent on all bit strings of length N.

Well, yeah...

[update()]

It's apparently much easier to compare against existing patents than sift through hundreds of surviving algorithms to determine if they are useful, original, and patentable.

Or, as the article puts it:

"I imagine we have done that but we don't know it," laughs Koza. To identify valuable, original results rather than simply matching patents, he explains, a human expert in the given field would need to evaluate tens of thousands of survivors.

In other words, it turns out that an infinite number of monkeys really will stumble into everything given infinite time. Frequently, solutions are simple -- it's identifying the need and fitting a solution to it that' s worthy of a patent.

Banalities

[Blackheart2]

The fact that a machine can come up with patentable software only goes to show how banal most software patents are.

Re:By Definition, should not be Patentable.

[037]

Okay, listen. The Genetic Algorythm is NOT merely the act of analysing absolutely everything that is possible. That is not feasable. It is not iteration, either. The GA is a process that sets very specific 'fitness' functions and models a limited set of variables that can be changed randomly in order to work towards the highest fitness possible. Each genetic (should be called evolutionary, but it seems that I have already lost that battle) program is usually tailored for a specific problem. If you think that you can write a program that models absolutely every possible invention, then go ahead and try. I have been working on a limited algorythm as a hobby project, and a decent fitness function is a very difficult creature to track down.

Re:We evolved it first! Nah-nah!

[rgmoore]

Here's a compromise for those who want to patent emergent solutions: useful GeneticAlg output should be allowed to be patented, but only if a human can also understand it; that includes the means and the ends. That's only fair, since patents were meant to reward human ingenuity in the first place, not some multi-million dollar darwin cluster breeding the shape of a square peg into a round hole.

This is wrong for several reasons. Patents were not invented to "reward human ingenuity"; they were invented to promote the development of useful arts. The original goal of patents was to encourage inventors to disclose their discoveries rather than keep them secret. The inventor was given a limited term monopoly as an inducement but at the price of publishing his methods. That way other people would eventually be able to benefit from the invention and it wouldn't be lost when its inventor died. Understanding exactly how it works is completely secondary. And, of course, publishing the idea is one very good way of helping to figure out how it works; even if the original inventor doesn't understand it, the chances are good that somebody else will be able to figure it out and develop it further. That's an argument in favor of allowing patents of non-understood ideas.

More importantly, requiring that software patents be understood applies a much stronger standard to them than other fields. People in other fields patent things that they have little or no coherent understanding of all the time in other fields. Chemical engineering is a great example of this. There are lots and lots of chemical processes that are understood poorly if at all. I'm not sure if anyone really understands a lot of forms of heterogeneous catalysis, for instance, and most forms are certainly patented before people understand them clearly. People know as a matter of experience that following certain recipies will produce the results they want, but exactly how those procedures work is anyone's guess. Asking software to adhere to the higher standard that the inventor be able to explain exactly how it works is unfair.

That's not to say that I'd necessarily run right out and use an algorithm whose operating principles were not clearly understood! After all, if you're not really sure how it arrives at its results, you'll never be sure that it will always arrive at the right result. But you're unlikely to gain anything by preventing people from patenting non-understood algorithms. All you'll accomplish is to stifle the chance that somebody will learn how they work and thus how they can be made to work better.

By Definition, should not be Patentable.

[A Commentor]

So if a computer is able to determine the algorithm, How can it be argued that this is not something that someone trained in the art would not come up with?

If someone could program a computer to iterate over possible solutions, this SHOULD NOT meet the requirements for Patentability.