>1. You're basically patenting algorithms, which aren't supposed to be patentable.
Hardware patents just cover algorithms that have steps involving arranging atoms.
>2. There seems to be an assumption that anything new is unobvious and therefore patentable. And in computing that isn't really the case.
It isn't really the case in anything. That no one has bothered to patent yellow wrenches with beveled edges doesn't mean they should be patentable. That the software equivalent is currently more likely to be granted isn't really an issue with the idea of patenting software.
>A patent is supposed to describe how to actually do the thing. Software patents by and large don't do that.
That's an issue with particular software patents, not one particular to patenting software.
>hardware patents just cover algorithms that have steps involving arranging atoms.
Yes. Further steps have been done. Its the difference between copywriting the idea for a book, and copywriting an actual book.
>That the software equivalent is currently more likely to be granted isn't really an issue with the idea of patenting software
It's an issue with the current incarnation of software patents.
That's like saying IC cars aren't bad for the environment because we 'could' fuel them all with biofuels and have a carbon capture thing on the exhaust.
In the real world an IC car can rightfully be criticised for being bad for the environment.
>That's an issue with particular software patents, not one particular to patenting software.
Again, the issue doesn't have to be inherent to be valid.
Hardware patents just cover algorithms that have steps involving arranging atoms.
>2. There seems to be an assumption that anything new is unobvious and therefore patentable. And in computing that isn't really the case.
It isn't really the case in anything. That no one has bothered to patent yellow wrenches with beveled edges doesn't mean they should be patentable. That the software equivalent is currently more likely to be granted isn't really an issue with the idea of patenting software.
>A patent is supposed to describe how to actually do the thing. Software patents by and large don't do that.
That's an issue with particular software patents, not one particular to patenting software.