A "For the Record" column of errata will run under the editorials whenever it's appropriate. The first one appears today. ....
There is another, Godzilla from hell kind of correction that generally requires lengthy explanation....
A classic case of correction run amok involved a column that Paul Krugman wrote on Aug. 19 about the Florida recount in 2000 in which he said that two different news media groups reviewed the ballots and found that "a full manual recount would have given the election to Mr. Gore." That was incorrect. Paul tried to clarify things in his next column, but the public editor, Byron Calame, objected that since nothing in the second column was labeled a correction, the original error would survive in the permanent record.
Paul published a correction in his next column. Unfortunately, the correction was based on information published in The Miami Herald that was wrong and had never been formally fixed. Paul appended another correction to the Web version of his column, but asked if he could refrain from revisiting the subject yet again in print.
I agreed, feeling we had reached the point of cruelty to readers. But I was wrong. The correction should have run in the same newspaper where the original error and all its little offspring had appeared. Here it is:
In describing the results of the ballot study by the group led by The Miami Herald in his column of Aug. 26, Paul Krugman relied on the Herald report, which listed only three hypothetical statewide recounts, two of which went to Al Gore. There was, however, a fourth recount, which would have gone to George W. Bush. In this case, the two stricter-standard recounts went to Mr. Bush. A later study, by a group that included The New York Times, used two methods to count ballots: relying on the judgment of a majority of those examining each ballot, or requiring unanimity. Mr. Gore lost one hypothetical recount on the unanimity basis.