Shortening the patent period (he suggests 3-5 years) will go a long, long way to addressing many of the current problems.
It would incent companies to focus on core inventions, instead of filing 20-year lottery tickets.
Patent value to trolls would be greatly reduced, lowering the incentive for nuisance lawsuits. (Companies can continue to file new patents as they continue to innovate).
It would increase the incentive for universities to commercialize ("get moving before the patent expires!").
In Jefferson's day, when information and innovation moved much more slowly, a 14yr patent period (now 20 in the US) made more sense. Today, it just means the set of in-force patents diverges wildly from the leading edge of innovation.
For some context, Bezos wrote this letter months after Amazon was awarded its infamous patent on one-click checkout. Amazon used the patent to sue Barnes & Noble and ultimately force it to add extra steps to the checkout on barnesandnoble.com. Bezos wrote this letter a few days after a scathing op-ed about the patent and the lawsuit came out in the NYT, but never chose to turn the patent over to the public domain. It's now been 20 years and the patent is expiring next month.
If done right -- and it could take 2 years or more -- we'll end up with a patent system that produces fewer patents (fewer people will bother to apply for 3 or 5 year patents, and fewer patents means less work for the overworked Patent and Trademark Office), fewer bad patents (because of the pre-issuance comment period), and even the good patents won't last longer than is necessary to give the innovator a reasonable return (at Internet speed, you don't need 17 years).
There is very little in the way of empirical research on the effectiveness of patents and specific patent policy elements, so regulators are fairly hamstrung; they know the area is important but mis-formed, but they can't really get some form of consensus as to which elements are broken and which aren't.
In Canada, for instance, federal government's patent policy research apparatus is essentially two people. Two.
Additionally, the epistemic academic work required to provide a proper critique of the industry is not well funded by public or private sources. On the private side, most interested parties are incentivized to be patent maximalists, while on the public side, applicants are basically pitching an argument to the legal academy that indicates that their central dogmas are so tremendously wrong so as to be unjust.
Additionally, the issues caused by a lack of patent reform are not acutely felt by the majority of individuals or industries, which means the benefit is concentrated while the pain is distributed. Representation at industry forums, accordingly, is slanted.
In the pharmaceutical space, for instance, WIPO had no issue with crushing access to medicine until there was a global outcry, and patted itself on the back when they introduced access provisions, despite themselves admitting the provisions were basically unused after four years following their introduction because the local industries and talent that they presumed were already dead and gone.
I'm not so sure it's so clear cut. Some inventions genuinely take years to get ready (cough cough anything semiconductor) and even small things require significantly longer time periods when the filer doesn't have a whole lot of money, eg is a startup or a lone wolf.
That being said, the system is clearly broken.
Two suggestions i have is to shorten software patents to 10 years and adjust the length of a patent based on the financial capability of the filer (eg pegged to inflation). The other thing is that filing for a patent should not be so expensive because of the lawyer, it's just ludicrous. If the lawyer can be removed via a combination of software and policy changes (either pubic efforts, private efforts or both), then people will be able to pay significantly more for examination. For small inventors, a "slow lane" that is significantly cheaper can be allocated. Another idea is anonymous (for the filer) crowd sourced prior art search with incentives.
I know there are much more pressing problems than patent reform but it's too bad no changes (that I know of) have been made in the 17 years since this was written.
It would incent companies to focus on core inventions, instead of filing 20-year lottery tickets.
Patent value to trolls would be greatly reduced, lowering the incentive for nuisance lawsuits. (Companies can continue to file new patents as they continue to innovate).
It would increase the incentive for universities to commercialize ("get moving before the patent expires!").
In Jefferson's day, when information and innovation moved much more slowly, a 14yr patent period (now 20 in the US) made more sense. Today, it just means the set of in-force patents diverges wildly from the leading edge of innovation.