the ALGORITHM patent *IS* "To catch a mice"
Exactly my point. _THAT_ kind of patent (algorithm or otherwise) is horrible, because it covers way more than it should. The fact that the USPTO has been allowing those kind of patents means that the _current_ patent system is broken. It doesn't necessarily follow that the _idea_ of patents is bad.
And what algorithm is it anyway?
The quicksort algorithm. If you've been a programmer for 23 years you should recognize that that is a well-known, defined name for a specific algorithm (as opposed to say a bubble sort or a merge sort or any number of other sorting algorithms.) Its not _just_ "sort quickly," but a specific method for doing so.
Someone can come along with a spring loaded mousetrap that uses a different trap door form and be a different patent.
Depends how different it is. If its the exact same mechanism but closes from the left instead of the right or uses a 3" bolt instead of a 2.5" bolt but otherwise identical, you're probably not going to get a new patent granted even if it doesn't _exactly_ match the original patent's blueprints.
No it didn't. The blueprints were used as a reference, not as the only possible construction covered.
any reasonable analogue of a patent of software would be "the code"
Now that actually brings up an interesting idea -- require a pseudocode "blueprint" to be included in the patent description. Not sure entirely how much that will solve (since the worst software patents are pretty easy to duplicate anyway even without original source) but it might give companies a slight bit more pause before filing a junk patent, if for no other reason than someone will have to spend the time distilling their code down to the core algorithm.
But they need to be proper patents, not "Catch a mouse" patenting of the goal of the software
Which is exactly what I've been saying. I'm not sure why you're sounding like this is contradicting my previous statements.
surely that would be "better code"
By what reasoning? Chances are someone who spent a year coming up with perfecting an algorithm to patent is probably going to have written a better implementation than some guy who goes "that's a good idea" and hacks some shit together that does basically the same thing. For a real world example.. I'm assuming you know what a mouse trap looks like and how it works. Could you go out and build one from scratch that works as well as the ones built by a manufacturer who has been building them for 50 years? Probably not.
least DIFFERENT code, therefore would get a DIFFERENT patent
Only if its different enough that you can justify it not being covered by the existing patent. Writing essentially the same algorithm in BASIC where the original implementation was in C++ isn't exactly what anyone (with a brain, which doesn't necessarily include USPTO) would consider "novel."
they don't require invention to make
Come again? You've never had to stop and spend half a day thinking about what approach you need to take to solve a particular problem? You're either the most godlike programmer that ever existed, or your claim of 23 years programming experience is going to get called into question.
And anything patentable (legitimately patentable not "does something with a computer" type crap) would likely take far far longer than half a day to invent, distill down into a core algorithm, prove its complexity category, etc. I suspect a simple algorithm meeting all that criteria would take at least 6 months to file properly, not even counting delays imposed by USPTO's red tape.
they require invention to IMPLEMENT
So if I dream up a new mouse trap, jot down some blueprints and file an application without ever building one, I'm in the clear but if I dream up a new sorting algorithm, jot down the generalized procedure without ever writing real code, I haven't actually done anything? In both cases I've spent time thinking up and developing my idea, and in both cases I have no idea if it will actually work since I never built one.. but somehow the mouse trap patent is more valid in your opinion?
Keep in mind, there is _nothing_ stopping the USPTO from granting a patent on a super-vague "a mouse trap that traps mice" type patent for a physical object. The only reason they don't accept patents like that in the real world while accepting them left and right in the software world is that they understand real devices better than they understand software algorithms. Again, its the _system_ that's broken. Not the _idea_.
I certainly have strong doubts whether the system will ever transform into something that isn't broken (as is common when discussing fixing political systems, the people most in position to fix things are frequently the same people who most benefit from keeping them broken,) but abandoning the system completely is even more unlikely than reforming it so suggesting plausible improvements is still more useful than shouting "burn it all!" even if you truly believe they should be completely done away with.