"Accelerating innovation" is something which, if it exists, touches only a few areas of engineering. In most areas of engineering, innovation has gotten more expensive, not less. Very few things are amenable to a distributed group of hackers working on it. Jet engines are still best designed by a roomful of PhD's. And those cost lots of money.

We can argue about how necessary patents ever were, but I don't think they're any less necessary today.

The question, in my opinion, isn't whether patents are still relevant. Its whether we can draw lines better to leave out less useful ones while keeping that exception from swallowing the patent system. E.g. if a power control module for an LTE radio is patentable if implemented in an ASIC, should it become unpatentable when you implement it in a DSP? If not: how do you allow those patents while invalidating Amazon's One Click shopping patent? To a certain extent, its tempting to just say: "its too hard to draw the lines, its not worth the trouble." That's easy to say if you're not Qualcomm and don't have a dozen companies who'd love to use your work for free.

It's worth drawing a line between "a particular solution to the problem of power control on LTE radios" and "any method of doing power control on LTE radios". Far too many software patents seem to cover any solution to the same problem, not just a particular implementation.

(Note that the software in the DSP also gets the benefits of copyright protection)

Possibly the wrong example for your argument :) Jet engines can now be designed by a team at SpaceX rather than at NASA, and the team can 3D-print them. That seems like reduced-cost innovation compared to decades past.

http://www.fool.com/investing/general/2014/06/02/elon-musk-m...

(PS - I definitely don't disagree with your general point, but it might be becoming harder to know which areas need patents and which don't.)