Unsurprisingly, I have OPINIONS about this, starting with:
Either you think Church-Turing (*) is wrong, or patents are the only currently sensible IP protection for software programs. If you can't differentiate a program implemented in software from the same program implemented in an EPROM from the same program implemented as a custom chip (and I don't think anyone can from the outside) then a software program is a machine and patents are how we protect machines.
THAT said, the current state of software patenting is a giant frelling mess. I could go on about the reasons, and fixing some of them would fix a lot of the problems with software patents. E.g. immediately reject any software patent with no non-patent prior art - compare patents in software with, say, patents in bio. The latter are remarkably well-sourced.
I would also support a compulsory licensing regime for software patents, and make it particularly apply to any patents relevant to industry standards as determined by bodies like W3C.
tl;dr the problem isn't patents; it's that software patents are being managed by a bunch of greedy incompetents, not least of them giant companies like IBM that have done a good job of regulatory capture for their own commercial advantage. The fact that one can become a patent examiner in the software section without even the slightest knowledge of coding is an example of this.
(*) for the non-geeks, the Church-Turing thesis claims, effectively, that anything you can compute (i.e. write as a program) is mathematically identical to a machine of a certain type.
no subject
Either you think Church-Turing (*) is wrong, or patents are the only currently sensible IP protection for software programs. If you can't differentiate a program implemented in software from the same program implemented in an EPROM from the same program implemented as a custom chip (and I don't think anyone can from the outside) then a software program is a machine and patents are how we protect machines.
THAT said, the current state of software patenting is a giant frelling mess. I could go on about the reasons, and fixing some of them would fix a lot of the problems with software patents. E.g. immediately reject any software patent with no non-patent prior art - compare patents in software with, say, patents in bio. The latter are remarkably well-sourced.
I would also support a compulsory licensing regime for software patents, and make it particularly apply to any patents relevant to industry standards as determined by bodies like W3C.
tl;dr the problem isn't patents; it's that software patents are being managed by a bunch of greedy incompetents, not least of them giant companies like IBM that have done a good job of regulatory capture for their own commercial advantage. The fact that one can become a patent examiner in the software section without even the slightest knowledge of coding is an example of this.
(*) for the non-geeks, the Church-Turing thesis claims, effectively, that anything you can compute (i.e. write as a program) is mathematically identical to a machine of a certain type.