Aleks Yankelevich drew my attention to an interesting patent case
Gore versus Garlock. The Federal Circuit court decided
to reverse a lower court reaching the stunning conclusion that prior art doesn't matter if it was kept secret. So if you invent something and keep it secret, someone else can patent it.
To be clear: in terms of promoting innovation - if someone already invented it once without claiming a patent that's pretty good evidence that no patent was needed to provide the incentive to innovate. There is the argument that one of the purposes of patent law is to give incentive to reveal secrets - the problem is that argument doesn't make sense. If patents last 20 years I'll only take out the patent if I don't think I can keep the secret for 20 years - in which case nobody will be able to use the information I revealed until after they would have figured it out anyway.
The ability of Congress to ignore any evidence (see SOPA/PIPA) and the courts to ignore common sense is frightening.
A few points & clarifications with respect to your post.
I disagree with the ruling because the previous inventor, who was apparently keeping their invention as a trade secret, is now in a legal bind. They have been practicing the invention for a significant length of time before a subsequent inventor made the same invention. The risk of trade secrets has always been that a later inventor can effectively remove the trade secret by a parallel invention, but jurisprudence has previously held that prior use of a trade secret prevents later patenting. Since the patent has issued and has been upheld, can the later patenter sue the company that had it as a trade secret? There is no logic to that.
Does that also mean that a company with a trade secret they have been practicing for some period, say 20 years, now decides to patent their trade secret. They may decide to do so because a competitor appears to be close to coming up with the same invention.
For these two reasons alone, this ruling should be overturned if it is appealed, and I hope it will.
The principal point of patents is speed. Let's say a company invents something, but does not plan to use it for a decade. If they squirrel their invention away in the closet, they risk someone else inventing it some time before they plan to use it, meaning they lose their competitive advantage. However, the moment they reveal their invention, others can immediately use the teaching of that invention to develop a comparable technique that invents around the issued patent, if there is sufficient economic incentive. Lots of companies do exactly that.
Patents do not last for 20 years. They last for 20 years from the date of filing. Because of the time it takes for examination, the average patent has a potential lifetime of 15-17 years. I say potential because patents expire roughly in thirds because maintenance fees are not paid to keep them maintained. Thus, about 30% of all patents expire four years after issuance. About 60% of all patents expire 12 years after issuance, and the remainder, which is less than 40% of patents, go to term. The average lifetime for all patents is thus somewhere between 8 and 10 years.
