The AIG Life Insurance Numbers

The document itself is dated February 26, and Andrew Ross Sorkin's column on it came out on March 3, under the headline "The Compelling Case for Saving A.I.G., by A.I.G." He begins:

Inside the corridors of power in Washington, a 21-page document has been getting a lot of attention. It is marked confidential and titled "A.I.G.: Is the Risk Systemic?"

Sorkin devoted his entire column to the contents of the document, and wrote this:

In the United States, A.I.G. has more than 375 million policies with a face value of $19 trillion. If policyholders lost faith in A.I.G. and rushed to cash in their policies all at once, the entire insurance industry could falter.

I picked up that startling number on March 3, and then dropped it on March 4, saying that it was clearly false and that there should be a correction on Sorkin's column.

On March 9, Calculated Risk made the document public. At that point, everybody could see what it actually said, on page 9:

AIG has written more than 81 million life insurance policies to individuals worldwide
- Face value: $1.9 trillion

On March 10, the correction finally got appended to the bottom of Sorkin's column:

The DealBook column last Tuesday, about the systemic risks posed by any collapse of the American International Group, misstated the size of A.I.G.'s life insurance business. The company has more than 81 million life insurance policies with a face value of $1.9 trillion globally, and says that a run by its policyholders to cash in policies with cash value could result in the collapse of the entire life insurance industry. A.I.G. does not have more than 375 million policies with a face value of $19 trillion; that is the total of all policies held by insurance companies in the United States.

It strikes me as a little odd that the correction didn't appear until after the document was made public, when the document is crystal-clear about both numbers: if Sorkin actually had the document, it's hard to see how he would have made this mistake in the first place, and even harder to see why he wouldn't have corrected it immediately, after simply looking at the document to see if it really said what he said it said.

So did Sorkin write an entire column about a document he didn't actually have posession of? Or did he just think that any correction could wait a week until the next time his column appeared?