It's commonly brought up that the real problem that has led us into this enormous software patent mess is that most of these patents are on things that are obvious (and yet the patent office has granted them anyway). I would agree with this, and in fact, I would take it further.
There are different levels of obviousness. Something can be actually obvious beforehand[0]. Or, it can be merely retrospectively obvious -- once you see how it's done, it's obvious. Of course the latter is far too broad, and arguably includes nearly, well, everything.
But something can also be what I'll call semi-retrospectively obvious: It's obvious how to do it if you merely know that it's been done, but not necessarily how. Now, normally when people speak of something being obvious, they mean it was actually obvious, i.e. obvious beforehand, but I think the appropriate level of obviousness to apply to patents is actually semi-retrospective obviousness.
The reason is that the point of a patent is to encourage inventors to be open about the workings of their inventions rather than hiding them (and in exchange they get a temporary limited monopoly). But if something is semi-retrospectively obvious, there's already no incentive for inventors to obscure the workings of their products, since everyone else will quickly figure it out as soon as they know it's been done. I'm not saying such inventions are worthless -- often very important things are semi-retrospectively obvious -- but they do not really fall under the stated intent of the patent system. (But then, neither do design patents, which indeed I think all fall under this category, so...?)
And while people can argue endlessly about just what is and is not obvious, it seems pretty clear that most of the software patents that are causing this mess are, at the very least, semi-retrospectively obvious. Most computer programs do not require any sort of new algorithmic insight to write; programming is generally hard because the things you're writing are damn complicated, not because you're right up the edge of human knowledge. Truly new algorithms, well, maybe those deserve to be patentable. Though, y'know, some sort of compulsory licensing would be nice.
-Harry
[0]To someone skilled in the art -- the patent office doesn't seem to be applying that bit.
There are different levels of obviousness. Something can be actually obvious beforehand[0]. Or, it can be merely retrospectively obvious -- once you see how it's done, it's obvious. Of course the latter is far too broad, and arguably includes nearly, well, everything.
But something can also be what I'll call semi-retrospectively obvious: It's obvious how to do it if you merely know that it's been done, but not necessarily how. Now, normally when people speak of something being obvious, they mean it was actually obvious, i.e. obvious beforehand, but I think the appropriate level of obviousness to apply to patents is actually semi-retrospective obviousness.
The reason is that the point of a patent is to encourage inventors to be open about the workings of their inventions rather than hiding them (and in exchange they get a temporary limited monopoly). But if something is semi-retrospectively obvious, there's already no incentive for inventors to obscure the workings of their products, since everyone else will quickly figure it out as soon as they know it's been done. I'm not saying such inventions are worthless -- often very important things are semi-retrospectively obvious -- but they do not really fall under the stated intent of the patent system. (But then, neither do design patents, which indeed I think all fall under this category, so...?)
And while people can argue endlessly about just what is and is not obvious, it seems pretty clear that most of the software patents that are causing this mess are, at the very least, semi-retrospectively obvious. Most computer programs do not require any sort of new algorithmic insight to write; programming is generally hard because the things you're writing are damn complicated, not because you're right up the edge of human knowledge. Truly new algorithms, well, maybe those deserve to be patentable. Though, y'know, some sort of compulsory licensing would be nice.
-Harry
[0]To someone skilled in the art -- the patent office doesn't seem to be applying that bit.
