I had an interesting discussion with a friend of mine who has two PhD's (one in math, one in electrical engineering). He says way too many papers are published these days. Looking at the archive (xxx.lanl.gov) I would tend to agree. The idea of publish or perish combined with post-doc absurdity is putting too much pressure on people to write papers. The result is that too many papers are being published, and physics is by far the leading offender. In the old days a result had to be pretty significant to warrant a research paper. These days that isn't the case. As fast as a geek can type a new paper appears on the archive. It would be an interesting exercise to find out not only how many papers are really frivolous, but how many dupilicate other crap on some level that has already been published. String theory and "quantum information/quantum computing" probably lead the way for over production of papers.
I was also thinking about the Nobel prize. Maybe we shouldn't give one every single year. After all the Fields medal is only given out every 4 years. The Nobel prize has such an air about it, yet we have to give one year in and year out regardless of whether Nobel quality work is really being done. How many of the Nobel prize winners in the last ten years are really equivalent to say, Albert Einstein or Marie Curie? Probably none. Yet by giving a Nobel prize every year, we give many people an automatic stamp of genius that many don't quite deserve. It will never happen, but I think the scientific world would be better off if Nobel prizes were cut back to one every four years.
And stop publishing so many papers! Every calculation you do isn't worth writing up.