The value of protecting algorithms

When I worked in Indigenous affairs, I posted my thoughts on that. Well now it's time for my new work to inspire some posts. Certainly, I'm not the first person to think about these things. And definitely, some of my friends have thought these things through much more thoroughly than I have, so I welcome your comments.

Working in the school of ITEE means I cover computer science as well. This means that a vast majority of the ideas that come across my desk are related to algorithms. Cool, right? But I'm begun to wonder, what is the value in protecting algorithms?

A researcher comes up with a really cool way to do something, say, searching video databases. The way he does it is different to his colleagues and works just as well. Now he's gonna publish a paper on this, cause that's what he's paid to do. But as soon as he publishes, the method is in the public domain, which in most countries means that you can't patent it. So, before he publishes, we rush around to get a provisional patent put in to cover the space. Phew, right, now we've got the beginnings of protection for this algorithm. He publishes, and further down the track, the patent is published as well. This means that no one else can use this algorithm without his permission, right?

But here's the rub, 1) algorithms are easy to tweak, tweak a bit here, a bit there, maybe you're no longer infringing and 2) how do you enforce it? I can suspect that Google's YouTube is using this algorithm, but I don't really know, and without being Viacom, I don't think I have much of a chance of getting Google to reveal their algorithm to me. So, why patent in the first place?

Counterpoint: Google patented pagerank (the secret sauce behind why their search engine dominates). Do they sue people they suspect of using it. Does yahoo really not use some approximation of pagerank?

Thoughts?

Next time: When is it diagnosing a problem, when is it just dobbing colleagues.