Last week Salon.com, a publication I helped edit for many years, officially retracted “Deadly Immunity,” a 2005 story by Robert F. Kennedy Jr. that had promoted scientific research (never very persuasive and now widely discredited) linking autism to preservatives once used in vaccines.

It’s not often that publications go back and fix this sort of problem. It’s painful to admit error; the public rarely keeps track of past sins; and journalists generally work facing forward. Every now and then you do encounter massive cleanups of the record like this one, which ran in a Kentucky paper in 2004: “It has come to the editor’s attention that the Herald-Leader neglected to cover the civil rights movement. We regret the omission.” More often, we just let the imperfect record stand.

So here’s one cheer to Salon for bothering to deal with this issue at all. The occasion was the publication of Seth Mnookin’s new book, “The Panic Virus,” which takes the media — Salon included — to task for its part in spreading vaccines-cause-autism misinformation. (I was away writing a book in 2005 when Salon posted Kennedy’s piece, so I wasn’t involved in that decision and don’t have any inside perspective on it.)

A second cheer to Salon for actually owning up to the retraction in a forthright and prominent way, rather than simply burying it. That’s what the editors apparently did at Rolling Stone, which had also published Kennedy’s story in a joint publishing deal: The story’s gone from their website with no explanation or discussion.

To Delete or Not?

But I’m going to withhold cheer number three for my former colleagues, and here’s why: The site, which had already run a slew of smaller corrections about the story back in 2005, decided to delete the article completely from its archives. Of course, you can still find it at the Internet Archive, and Kennedy has a version of it posted on his own site. But it’s gone from Salon. As editor Kerry Lauerman wrote in the retraction post: “We’ve grown to believe the best reader service is to delete the piece entirely.”

On one hand, you can easily see why Salon chose this route. (It has done so at least once before.) I asked Lauerman to go into the reasoning behind the choice, and he emailed me this:

It’s a seriously flawed story we feared could do real harm. People who have bought into the anti-vaccine panic have created a health crisis, and a flawed report that feeds that hysteria poses a real threat. With this particular story, the unproven logic that animates the piece — as when Kennedy says “the link between thimerosal and the epidemic of childhood neurological disorders is real” — is not easily excisable, and no matter how many editor’s notes or Drudge-like, red-flashing sirens you place on a story to warn readers, there will be those who will take a well-known, respected American at his word. We simply didn’t think it was worth that risk.

What should editors do with stories that they’ve lost confidence in? What does a news organization do when it concludes, not that specific facts were in error, but that the entire premise of a significant story simply does not hold water?

Offline publications don’t face this correct-or-delete dilemma — but there’s virtually no such thing as an all-offline publication any more, so this is everyone’s problem.

Though I understand Salon’s logic in choosing to delete, I’ve come to feel that the best practice in such cases is not to efface the record but to amend it. As original publisher of a story, you have the ultimate responsibility for maintaining the authoritative record of its history. If it’s been corrected, you can be the steward of that process. (As the invaluable Retraction Watch blog notes, the version of the autism story on Kennedy’s site doesn’t even reflect Salon’s 2005 corrections.) Annotating the story at its original URL maintains the integrity of the information, whereas deletion — with its Orwellian “memory-hole” overtones — opens the door for distrust and conspiracy-theorizing.

But what about your responsibility to avoid “real harm”? If bad information is a clear and present danger, aren’t we obligated to remove it? Maybe. But isn’t keeping misinformation in view and labeling it as such a sharper, more resonant public act?

Other Options

There are all sorts of tactics one might use to insure that nobody could read a story like “Deadly Immunity” without knowing that it has been retracted: Popup notifications, “interstitial” messages timed to precede a page, pass-through pages requiring a click to show you’ve read the notice — the entire arsenal of commercial Web publishing, usually directed at the mundane needs of advertisers, could be trained on this problem. Today this sort of stuff is rarely supported by our publishing systems and requires some extra technical work. But it’s all feasible. (On Twitter, which increasingly serves as a news network, it’s much harder to tie a correction to an original erroneous message — yet even there I think the best practice is to correct rather than delete.)

What’s the payoff of such extra effort? Preserving and presenting the entire saga of a story-gone-bad gives the public a chance to understand the full arc of a controversy. It also reaffirms the journalist’s primary mission to illuminate true stories rather than hide problems. The next time a publication faces a dilemma like Salon’s, I hope they’ll choose to make that effort.