Krugman on Social Security again

Although my shock and awe about the Bush Administration's fecklessness has been pushing me ever further into Krugman's camp, I still can't join better-known bloggers such as Josh Marshalland Brad DeLong in reverently embracing everything he says about Social Security.
Case in point: Krugman's column today.
The big point he makes is that the Bush plan probably amounts to nothing more or less than "government borrowing to speculate on stocks." This is probably correct, although we haven't seen the plan yet.
So maybe I am just quarreling over the footnotes, but they are important.
I have been saying that an individual accounts plan actually could eliminate the Social Security portion of the fiscal gap IF - as I do not expect for a second - it cut guaranteed benefits by the amount of the diverted payroll taxes PLUS $10 trillion in present value terms.
For Krugman, even this would not be good enough. This difference of opinion reflects that, while I am a fan of long-term measures such as the fiscal gap and generational accounting, he seems to think that, since the good ol' deficit is the measure we have always used, it's jolly well good enough for him.
He makes the valid point that long-term measures rely on projections of future policy that may not be credible. For example, the bond markets might (rightly) not believe that a $1 trillion diversion of payroll tax revenues today was really, truly going to be offset by an $11 trillion, or even a $1 trillion, reduction of benefits in the future. Who knows what Congress will really end up doing down the road?
This type of problem has led me to the conclusion that, while the fiscal gap is analytically a coherent measure and the deficit is not, in the world of official budget measures you need both. Policymakers should be generally constrained (at least in the sense of an accepted policy norm) against increasing EITHER the deficit or the fiscal gap, lest they game the former through smoke & mirrors timing games, or game the latter through non-credible claims about policy in the distant future. (An example would be purporting to "pay" for individual accounts through a $1 million a person annual head tax, to take effect in the year 2050.)
With the Bush Social Security plan, of course, the measurement issue may boil down to whether they can ignore making the short-term deficit worse because the long-term picture ostensibly remains the same. Ignoring that it actually will have gotten worse if the future medicine is politically less credible than the current candy. And ignoring as well that the idea was supposed to be improving the long-term picture, rather than placing a bet on stocks while it ostensibly remained the same.