If I’d sketched out the current scenario last summer — computer systems don’t work for months, millions lose insurance, and by the beginning of December, only 1.2 million people have picked up coverage from the exchanges and Medicare combined — the law’s supporters would have rolled their eyes and shaken their heads at the wishful thinking of the law’s critics. And now they generally assume that it will of course get better — that by March 31, if not sooner, we will see a measurable and substantial reduction in the number of uninsured.

But while that’s certainly very possible, it doesn’t exactly seem inevitable. To be sure, I myself find it hard to believe that the number of uninsured people will actually rise, even temporarily, as a result of the law. On the other hand, the administration has been pretty quick to leak whenever they had good enrollment numbers, and we haven’t heard a peep since the beginning of the month. So however incredible, it’s at least a real possibility that we’ll see a net decline in coverage on Jan. 1 — or even on April 1.

That has implications beyond just the people who would be uninsured next year. A lot of these insurance pools are already pretty small. If enrollment is much lower than expected, many markets may not have enough customers to make a viable pool. Even if the insurance pool isn’t older and sicker than expected, a pool that is too small can be wiped out by a couple of unlucky and expensive illnesses.