I think you're absolutely correct in the first part of what you said, which is why the second part makes no sense to me. You said:
> I create a machine to do something in a new and novel fashion. I do it using mechanical parts and gears (which are an implementation detail -
> the individual gears and such aren't really new or novel, just the way the machine works overall). I should be able to get a patent for that.
> Now, I rip out the gears and put in motors and software to do the same thing the gears did
It's the invention itself that's patentable. Whether you build a particular version of the invention from gears (metal) or Gears (software) makes no difference -
you might well do both. You could have a crank up gear version that sells for $3 and an electronic version of the same invention for $30. It's the invention
that's patented, not a particular build of it. We agree on that, right?
So why a whole new class of IP separate from patents. Patents cover "useful things". Programs are useful things. I don't see any reason to treat things differently whether they are built of wood, silicone, or silicon.
I do also agree on shorter patent terms, by the way, because the rate of technological change has increased. Technology didn't change nearly much from 1880 - 1900 as it did from 1980 - 2000. Not just with computers - biomedical technology, energy technologies, we're in a faster paced world now, so faster paced patent expiration makes sense. (Also MUCH faster copyright expiration.)