I've read a few, admittedly mostly the infamous patents, and have yet to find anything useful. At worst, we have someone patenting doubly-linked lists (What if, instead of one pointer, lists had two pointers... and they could point anywhere! Or maybe they could have N pointers! Genius!) and other obvious dreck (A stick as a dog toy? Defeated only because two other people already had patents!?)

Or sometimes they file wish claims on things none of their products actually do. They don't require code to actually prove that you even could do what you claim to have invented, even though source code would be analogous to requiring blueprints for machines.

Anyhow, my point would be that I've never heard of anyone reading patents to learn anything about computer programming (maybe they do, but I've never heard of anyone learning that way). Maybe patents are a lot more useful with mechanics, I don't know, but the software patents I've seen are not even useless. Half of them read like they were written by the "I just need a typist, err, coder to implement this for me" people. I've even seen people say "you've got a flowchart, what the hell else do you need?"

Human AI Flowchart: [Evaluate state of world] -> [Think like a human] -> [Act] -> [Repeat]

(The similarity to the evaluate-apply loop in Lisp should be obvious.)

I'm not justifying patents. I also think they're generally not useful and destructive. But they are here.

It doesn't take much imagination to see why those are bad metrics for people more concerned with promoting the progress of science and useful arts.