I think the problem is basically unsolvable. The problem is that there are too many people who want to do research (even more now after the large increase in PhD funding in the early 2000s) who are competing for a very limited pool of research money. That's it, it's not changing, fooling around with the allocation is not going to fix it. I don't think it's a terrible idea to reduce MSTP slots somewhat and fund more R01s instead, but you'd have to cut three MSTP slots to fund one additional R01 so it would be a drop in the bucket given there are only about 300 MSTP slots funded by the NIH per year, compared to ~5000 R01s.
I definitely concur with the first sentence, given realistic constraints. The key issues:
1. Too many people go into science (PhDs), but there is not enough funding for them or their MSTP counterparts
2. It takes too long to get through training- by the end your priorities shift and money becomes a bigger concern. As a researcher, you will make less of it
3. There is little to no security in a science career right now
4. It is too difficult (or there is little incentive) for administrators to accurately predict which applicants will ultimately continue a research career
5. Too much lifestyle discrepancy between clinical and science careers
6. Reimbursement climate forcing young physician-scientists to commit to full-time clinical practice, even if it is not desired
I am a big proponent of cutting MSTP spots drastically to reflect the monies available on the back end (addressing #1). This will force medical schools to be more selective as to who enters their programs (addressing #4). The money generated from this cost-cutting measure should go to support MSTP graduates, with guaranteed monies for their start-up and salary support out of fellowship (addressing #3, #5 and #6).
I am not sure about today's numbers, but when I matriculated I got a letter telling me the total compensation of my training- it was ~$300K. This was in 1999, I bet it is a lot more today. A $500K payout to graduates for start-up out of fellowship would do the following:
1. Make them more likely to stay because they already have funds to work with
2. Make their departments more willing to give them protected time and be productive because they are paying for their own research time or reagents
3. Potentially allow MSTP grads to bypass the dreaded post-doc or even instructorship and go right into asst. Prof with a much more viable salary.