[whatshisface]

So basically the rules are, manufacturing in America is about lawyering patents, playing with fire, and being threatened by big companies who can bankrupt you whether you're copying their ideas or not. No wonder China is pulling ahead.

[ilyt]

Patent law was mediocre idea turned into atrocity by lawyers. It just stifles the progress at each step with being "beneficial" only to big enough corporations that can brute force the system with lawyers

[IshKebab]

I disagree. Patents are a good idea and mostly aren't that bad (you mainly hear about the bad ones).

This one seems like a solid idea worthy of some reward.

I think the patent system does need some tweaks but other than that it's not too bad overall:

* Software patents need a shorter term than 20 years. Maybe 10.

* It needs to be way easier to invalidate patents after they have been granted using prior art.

* Ideally they would simplify the crazy language in patents so they are easier to read. Do we really need to put "plurality" in every other sentence?

[ulnarkressty]

* The research was funded with public money, so everyone should benefit from the results.

* The one who did all the work (maybe even came up with the shape idea) was one of the interns. From my experience in industry and academia you will give away all your rights with your contract and get some pennies in return, while the department will get all the big bucks. Does it sound fair?

[ilyt]

Vast majority of software patents are maybe a month of work.

There is some reason to want something that takes years of research to be protected but basically nothing software is that.

> I disagree. Patents are a good idea and mostly aren't that bad (you mainly hear about the bad ones).

I'd wager if you removed bad and the trivial ones a lot of them would be gone.

Filter out ones that could be clean roomed by competent engineer in a week and 95% of them would be gone.

[frozenwind]

It'll get interesting once AI algos proliferate to other knowledge domains such as physics and engineering and start hallucinating solutions, as does GPT-3 now. Some of those solutions, adjusted and fine tuned, will have real world applications. How will patents be relevant then?