Usually when people say that, they mean either hiring more people or hiring better people by paying more.
Given the very low quality of granted patents, hiring more of same kind of people won't help. We'll clear backlog but since the percentage quality will remain the same, we'll just get even more bad patents.
Hiring better people seems more plausible but is also naive.
The biggest problem is that patent examiners have bad incentive: they are judged by number of patents accepted. Since a rejected patent can be re-submitted ad infinitum (after slight wording changes), their best strategy is to just accept a patent. I don't know of any negative consequence for patent examiner for just rubber-stamping bad patents (an example of that would be firing people after N patents they've accepted were found invalid by independent re-examination).
The second biggest problem is that the standard for a patent is vague and apparently very low. In theory patent should be novel and non-obvious, in practice (especially in software) they are ideas that a competent people come up with during regular work and not as a result of some year-long R&D process focused on one problem.
Finally, given the amount of patents it's absurd that anyone can actually do a fair job evaluating them. This is not something that scales by adding more bodies because in order to say whether a given patent application is novel, a patent examiner would have to do linear search of all existing patents to make sure that it's not like something that has already been patented (not to mention the prior art requirement). Even if we limit the amount of data by trying to search only on related topics, the search space is still absurdly large. It just cannot be done well, which is why it's done so poorly.
Also, patent claims are, as far as I can tell, nearly indecipherable. I worked on an application for an invention with really helpful, smart, technically literate lawyers, but after it was all put in the form of claims and figures and embodiments and specifications that could be (but are not necessarily?) implemented in digital electronic circuitry, I could barely it -- and it was my own work! Pity the patent examiners who have to make heads or tails of these things!
Usually when people say that, they mean either hiring more people or hiring better people by paying more.
Given the very low quality of granted patents, hiring more of same kind of people won't help. We'll clear backlog but since the percentage quality will remain the same, we'll just get even more bad patents.
Hiring better people seems more plausible but is also naive.
The biggest problem is that patent examiners have bad incentive: they are judged by number of patents accepted. Since a rejected patent can be re-submitted ad infinitum (after slight wording changes), their best strategy is to just accept a patent. I don't know of any negative consequence for patent examiner for just rubber-stamping bad patents (an example of that would be firing people after N patents they've accepted were found invalid by independent re-examination).
The second biggest problem is that the standard for a patent is vague and apparently very low. In theory patent should be novel and non-obvious, in practice (especially in software) they are ideas that a competent people come up with during regular work and not as a result of some year-long R&D process focused on one problem.
Finally, given the amount of patents it's absurd that anyone can actually do a fair job evaluating them. This is not something that scales by adding more bodies because in order to say whether a given patent application is novel, a patent examiner would have to do linear search of all existing patents to make sure that it's not like something that has already been patented (not to mention the prior art requirement). Even if we limit the amount of data by trying to search only on related topics, the search space is still absurdly large. It just cannot be done well, which is why it's done so poorly.