The bottom line is that if HSAs are a better deal for healthy people, then inevitably they're a worse deal for sick people. And if you take healthcare seriously, it's sick people you should be concerned about.
His argument is that with HSA's, healthy people can keep more of the money that they would otherwise devote to health care. But that means that the pool of people not electing HSA's will include fewer healthy people, and their insurance costs will go up.
To me, the "panacea" in health insurance--if there is one--is not HSA's per se. It is catastrophic health care coverage, instead of "insurance" that covers too much Grey-area medicine. Because HSA's tend to be complementary with catastrophic insurance (in part by legislative intent), and because I think that health expenditures among the elderly are predictable, I think that there may be a case for HSA's. What I mean by the latter argument is that all of us can expect to be sick as we age, so none of us is in the "healthy" pool from that standpoint.
Note that our family has an HSA with catastrophic coverage, and we are anything but healthy. So pulling us out of the "regular pool" is not raising anyone else's health care costs.
Catastrophic health insurance probably is a positive-sum reform. There is plenty of evidence that people are more conservative in their consumption of health care services when they share more of the costs.
What (potentially) sick people need is long-term catastrophic health insurance. HSA's do not provide that, but they are consistent with it.
Thanks to Tyler Cowen for the pointer, and for plugging my forthcoming book, which focuses on long-term catastrophic health insurance, not on HSA's.