Reuters to change how it handles retractions after killing David Cay Johnston’s News Corp. column

Reuters is planning to change how it deals with retractions on its website after killing a column written by David Cay Johnston which incorrectly stated that News Corp. had received billions in tax refunds.

When Reuters staff learned the day after it had been published that the column was wrong, they posted an advisory on Reuters.com saying the column had been “withdrawn.” The column itself, however, went untouched for another couple of hours, when it was removed. Links to the story resulted in 404 errors, and still do. That evening, Reuters posted a followup column by Johnston in which he acknowledged his error and described how it happened.

Here’s how Reuters will respond in the future, according to James Ledbetter, Op-Ed editor: A notice will be posted atop the offending article at the same time that the advisory goes out, and an editor will strike through the incorrect portion of the article. If another article is pending, as in this case, the notice will say so.

Johnston’s column illustrates a tough problem for news outlets that produce many stories a day: how to chase down misinformation once they hit “publish.” A post can be deleted, but no one can reach across the Internet and pull back every link, tweet and blog post.

In essence, Reuters will move from an approach that made sense when it was a wire service with no online publishing home to one commonly used by bloggers to update and correct their posts.

“The correction/kill policy that is followed at Reuters is long-established by the wire service,” Ledbetter said. “There isn’t a procedure for taking down something that is wrong because for the vast majority of Reuters’ existence, there was nothing to take down.”

Under the new policy, the erroneous post would remain online even after Reuters published a follow-up.

“I think it stands as a transparent record of what occurred,” he said. “I think to take it down – while I can see some argument for that – it’s not being fully transparent with our readers about the process, and it could be subject to abuse.”

One advantage of this approach is that it retains reader comments, which disappear when a post is deleted.

In this case, user comments could have helped Reuters and Johnston identify a critical error within a few hours of publishing the original story. Shortly after the column was posted the morning of July 12, a reader suggested in a comment that Johnston had misread News Corp.’s documents.

But that comment and another one that went into more detail were unnoticed, in part because Reuters doesn’t ask its writers to track story comments and in part because the story was posted in two places on Reuters’ site.

Johnston wrote in his retraction that the first clue that he was wrong came in the middle of the night on July 13, when he went to a tax policy blog and saw a comment pointing out his mistake.

Johnston posted a comment defending his conclusion. But it bothered him enough that he was up early the next morning checking his figures. That’s when someone with a tax policy lobbying group emailed to tell him his mistake.

The person who commented on TaxProf Blog and Reuters went out of his way to alert Johnston, posting on other blogs that had written about about News Corp.’s tax windfall. On July 13, the person posted again on Reuters, this time on another version of the story. Ledbetter saw that one, which he called “extremely perceptive,” and sent it to Johnston. By then, Johnston was already working on his mea culpa.

By the time that Johnston had tracked his error and a Reuters editor made the call to issue a “kill order,” the column had been up for more than a day and NPR had interviewed Johnston on “Morning Edition” to discuss his findings. (NPR brought Johnston back on the show on July 14 to explain his error.)

Told of how quickly this commenter had picked up the error, Ledbetter said, “We’re definitely reaching the right audience, and … we can learn a lot by creating more structured mechanisms between ourselves, our columnists and our commenters.”

Shortly after Reuters published Johnston’s column on News Corp., a reader said Johnston had misread the documents. A few hours later, the reader even pointed out how Johnston had made his mistake. Both comments went unnoticed, delaying the time it took for Johnston to realize that he was wrong.

Ledbetter said Reuters wants regular writers to consistently engage with commenters, and he’s planning to have regular bloggers moderate their own comments.

But considering that some writers contribute infrequently and they may not be comfortable with WordPress, he said, “we don’t really have as a matter of policy or practice any kind of rule or guidelines that comments need to be responded to.”

He said there hasn’t been any discussion of a “report an error” button that is in use on a few sites, though it could be useful.

Aside from the typical problems in discerning substantial error reports from inaccurate assertions, one reason this comment slipped by was that two versions of the story were posted online.

Stories that originate on the wire are automatically posted to Reuters.com, but editors publish another version on a WordPress portion of the site. That portion of the site has a better presentation and is organized by blog.

Normally, Ledbetter said, auto-generated posts get just a fraction of the page views. This was one of the few exceptions.

The new corrections policy that Ledbetter described won’t apply to the auto-generated posts because they can’t be changed. In case of substantial error, they’ll just be deleted.

We have made it easy to comment on posts, however we require civility and encourage full names to that end (first initial, last name is OK). Please read our guidelines here before commenting.