Anecdotal personal experience being part of patent applications has showed me that almost any idea, as long as there are no mainstream examples of prior art, can be patented given time (will take 4-5 years these days) and money. Most decently smart people will come up with patentable ideas all the time, just doing their job, but won't file because they don't have the desire or resources (and don't understand why their only somewhat-novel obvious-to-them idea is patentable).
"The idea that I can be presented with a problem, set out to logically solve it with the tools at hand, and wind up with a program that could not be legally used because someone else followed the same logical steps some years ago and filed for a patent on it is horrifying." - John Carmack
You'd almost think that they were getting money for accepting patent applications...
<Edit>This is a perfect example of mismatched incentives. The patent office has a lot of incentive (in the form of attracting user fees) for issuing dubious patents, and no real disincentive for doing the same.
Although, if I were to take your implication correctly - there should be no incentive to take a patent or reject a patent other than its merit.
Which would in turn imply no cost on applying for a patent, except its applicability after review.
But that would also create a perverse incentive for firms to submit patents all the time - there is no cost involved, no barrier of entry and hence no loss in making the effort.
Alternatively, we could have a very quick review system, which would mean that soon after patent submission you get rejection or acceptance. Which would mean that the patent office would need significantly more funding - considering the number of patents it receives vs people who have to review.
If there was a solution which could automate the search for prior art, that would be cool, and a way to reduce the size of the work load.
Its do-able, but I am certain that the law of unintended consequences was written to describe situations like these.
For example, in your suggestion, the part where we move the onus onto the courts, will gum up the courts. I live in India, where courts are constantly arbiting cases, and people know that if your case gets into court, it could be there for ever. Thats not a side effect we want to induce.
Now you could build in redundancy for that eventuality by expanding the number of people in court, justices and areas, but then in essence, you are moving the burden from department X to department N, with the added problem that those new people will be from law, and not a technical background.
I really do think that this is a case where people should just get someone whos a technocrat in charge, give him authority and funds, and then forget about it while the patent office is built back up into an institution that people respect.