Man, is everyone going at it hammer and tongs in the comments sections to the last two posts. There are a lot of good issues being raised, along with some interesting invective. But (as several people have noted), we're conflating several issues here.
On the subject of me-too drugs, I've already said a lot of what I have to say on the subject, and it can be found in that category over on the right. Looking over the comments, I tend to agree with SRC's take on the subject, although I'm pretty sure that my blood pressure doesn't go up as high as his does. A point that he makes, that many of these drugs are the sign of roughly simultaneous development (rather than sequential) is something a lot of people don't think about.
That said, there are some sequential me-toos, of varying degrees of utility. I've gone on about Clarinex/Claritin and Nexium/Prilosec here before, because I think that those are the two most egregious within-the-same-company examples. See this post on Sepracor for some others - come to think of it, they're involved in the Clarinex example, too, much to Schering-Plough's sorrow. It's fine with me if insurance companies decide that they don't want to pay for these things without a clear demonstration of medical need. If the market isn't big enough, companies won't develop such drugs. I'm in favor of as many pricing signals being sent as possible, all up and down the business.
Then there's the whole "how much does the federal research budget subsidize Big Pharma?" question. I'm going to have to pull that one out and deal with it - again - but not tonight. I continue to think that if you haven't been through the process, it can be difficult to realize just how long, twisty, and expensive the path between a great initial paper in Nature and a great drug on the market really is.
But these two issues have become thoroughly tangled up with the "can patents be improved" question. There have been some good suggestions scattered throughout the comments: longer granted terms for unmet medical needs, for example. If you try to apply these things too slickly, you open the way into terrible arguments, but for conditions where there isn't an existing therapy, the case should be clear. But isn't this exactly what the Orphan Drug Act is supposed to already do for us? There may not be as much uncaptured benefit here as we're thinking.
I wonder if there could be a (presumably shorter) fixed patent term that only starts once a compound comes to market, to deal with those long-gestation-period drugs that companies worry about getting involved with. You could also imagine adding onto a patent's term if the company agrees to do further studies (data to be made completely available,) which discover further utility (especially against competitor compounds?) But there's lots of arguing room in any scheme like that, too. And I could easily imagine companies carefully timing their investigations in order to stretch the patent out in the most lucrative way possible. (There was a similar suggestion in there for time limits to various milestones, tied to patent terms).
Coming up with incentives that can't be easily gamed isn't trivial, and it's a problem that's much larger than the patent debate.