[HarryHirsch]

what makes a "physical" patent you consider valid different from a software patent?

The investment. When you do process development or manufacturing the outlays for research and development are infinitely larger. You need a pilot plant, prototypes, testing, and soon you have spent tens of millions that you need to recover. In software you pay the salaries of a team of programmers. The investment in research that needs to be recovered before you can turn a profit is infinitely smaller, in software you have a government-protected money printing machine. Trade secrets work well enough.

[jacques_chester]

Patents are not granted according to development cost or profit potential.

[HarryHirsch]

That is the problem. You have people patenting the problem space, and then they just squat there, preventing innovation. Imagine NMR processing. One company patents fast fourier transform, the other patents entropy maximization. Everyone else gets to pay licenses. The power of government has been suborned for private profit with no public benefit.

[pnw_hazor]

You can't patent a problem space. You can get patents on innovations in a problem space.

Several patents from several inventors can be directed to solving the same exact problem. It is the 'how' that matters not the 'what'

[jacques_chester]

Patents are a tradeoff between granting a monopoly (generally negative) and creating a public good (generally positive). They're time limited for that reason.

If you find the patent valuable enough to license, you have the advantage that you have a description of the idea sufficiently detailed that you can recreate it after paying the license. The patent owner gets a return for their very chancy investment (well-prepared patents are expensive). After 20 years everyone can use it for free.

The alternative is that nobody shares anything with anyone. You still wind up with monopolies on ideas, but no incentive for anyone to convert their ideas into a public good after a fixed term.

[justinclift]

> The alternative is that nobody shares anything with anyone.

Open Source software demonstrates otherwise.

Also interesting, several major companies in existence today (including a well known chocolate manufacturer) were started when their home country decided patents weren't useful, so stopped using them.

[jacques_chester]

> Open Source software demonstrates otherwise.

Major opensource projects are predominantly developed by fulltime professionals working for companies that obtain patents as a matter of policy.

We have an example of what the world was like before patents: the guild system. It didn't appreciably advance technology.

[pnw_hazor]

Also, innovation is encouraged by forcing inventors to discover different ways to solve the same problem if licensing a patent is not an option.

[yxhuvud]

Knowledge would still spread as people change jobs etc. Your view of knowledge seem very company based, but the know-how resides in the people and they move around.

[jacques_chester]

Know-how is not the same as particular inventions. Besides which, confidentiality agreements would be much more aggressively enforced than they are now, to the point that they would make it unattractive to hire someone who's worked for a direct competitor.

[yxhuvud]

Know-how may not be the same as inventions, but it is also not a super relevant distinction. This is because know-how begets invention, as it leads to knowing what tools are available for solving problems, and also to knowing what problems needs solving. Removing patents would change some economics around, but in the end it will not change the primary driver towards improvement, which is that people want better and/or cheaper stuff. Companies don't need any other incentives to improve the stuff they do - if they don't, the customers will desert them.

Confidentiality agreements for employees and former employees that doesn't actually give the employees something in exchange for that confidentiality, is not legal in my jurisdiction.