Yup, Brad DeLong is right: I probably made a math error eight years ago, and didn't see it when it was initially pointed out to me by commenter Doug J. How large the error is depends on whether you take nominal or real GDP, compare it to nominal or real spending on the Iraq war, and so forth, but I'd say at a minimum I was off by a factor of at least three, though not "an order of magnitude", since it seems clear to me that nominal-to-nominal, not real-to-real, is the correct comparison.
I can't explain the math error, since it was, as I believe I mentioned, eight years ago. I can probably explain why I didn't see it, which is that I was laid flat with gastritis and working on about 3 hours sleep and 300 calories a day of rice and mashed potatoes. As anyone who has had gastritis can attest, this does leave you a little fuzzy. Or perhaps I was just being stupid.
I struggle daily to live up to Brad's well-known standards of accuracy, fairness, and integrity; I'm sorry that I clearly failed in this instance.