But that doesn't address my point, which was: who "establishes" that something is "a problem"? If you left it up to industry, CDs would never have been manufactured.
The patent clerk would read the published articles and make a judgement call. This isn't unprecedented; a patent clerks job is to make judgement calls.
What I'm suggesting is that this kind of judgement call is easier to make than the judgement calls we currently expect patent clerks to make..
In addition, this would leave a paper trail of the "proof" that the invention is non-obvious, allowing bad patents to be reviewed after they slip through the cracks.
You would have to have some kind of "official" method of establishing what was a "real" problem, which runs into the issue I mentioned above: some people would say it was a problem, some would vehemently deny it (and maybe even back their denial with lobby money). Result: innovative ideas never see the light of day.
Never see the light of day? Hardly. There's already precedent for making patent applications public after 18 months. The result would be more things in the public domain. It might bias the system toward having too many things in the public domain, but I think that stymies inventors less than the current system where too many things become patented.