One commenter on my earlier post snarkily referred me to Leonhardt's methodological explanation.  It says,

The stimulus adds $145 billion a year to the 2009-12 deficits. This number includes the additional tax revenue that the C.B.O. estimates will flow from the economic growth caused by the stimulus plan. This additional revenue amounts to about $40 billion a year.

Well, yes.  Standard Keynesian economics says that additional spending or tax cuts raise economic activity, leading to more tax revenue. But that economic analysis does not just apply to spending and taxes labeled as "stimulus." It applies to all the other fiscal policy measures as well.  It's really hard to defend making these adjustments for some fiscal decisions but not for others.

Leonhardt is usually reliable.  This looks like hack work.