This is probably the most cogent essay on software patents I've ever read, and it is a breath of fresh air when compared against RMS's nonsense.
My only gripe with it is the degree to which Graham thinks software patents are inherently okay. I buy all (or at least most) of his pragmatic arguments to the effect that patents actually have very little effect on modern software development. But that doesn't mean that the patent system is therefore good. Big companies tend to hoard patents in the way superpowers tend to hoard nuclear weapons; no one is stupid enough to actually use them. Okay, fine...but then why do we have patents at all? Graham argues that the USPTO is to blame for issuing idiotic patents. I agree. But I don't necessarily agree that it's possible even in an ideal world that the USPTO could differentiate between trivial and non-trivial patents. I mean really...how in the world would you codify what is a "trivial" software creation and what isn't? Some of the most powerful ideas in software are the simplest (think of, for example, auto-complete).
Oddly, Graham seems to indirectly recognize this when he talks about the different barriers to construction of real world inventions versus software inventions. Moreover, he (correctly) notes that the particular algorithms used _aren't_ what's hard about writing software. It's the thousand little details and performance optimizations that differentiate a good program from a bad one. Consequently, it can't possibly be patents that are creating the incentive for people to invent in the software world. It's the fundamental difficulty in copying the idea alone. Most of the time, the patent is irrelevent.
So, I dunno...the more I think about it, the more I am of the opinion that software patents are pointless. Copyright and trade secrets are enough to promote growth in the field.
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment