How news sites could improve accountability by tracking story changes, but probably won't
Web publishing has spawned a parlor game for media reporters, partisan bloggers and others who closely follow the news: finding stuff that's been deleted or changed on news sites and figuring out why. Points are awarded to the person who notices and cites exactly what was removed or changed. Bonus points for a screen shot.
The most recent game was played last month, when The New York Times removed a quotation by incoming Executive Editor Jill Abramson that said, in part, “In my house growing up, The Times substituted for religion.” In noting the deletion, some bloggers speculated whether the Times was trying to hide something, presumably Abramson's worship of false idols. The Times said the line was dropped for a fresher quote.
Although the deleted or changed material often isn't important outside the world of meta-media, sometimes regular news consumers become unwilling losers in this game. A witness' account in a developing news story disappears when it's replaced by the version that appears in the next day's paper. You remember a story saying one thing – and maybe you blogged about it – but now it says something else.
Scott Rosenberg, one of journalism's accuracy crusaders, has a solution: Have news sites save and publish previous versions of stories to enable people to see what was changed from one version to another.
A “versioning system” wouldn't affect the appearance of articles for most readers, who simply want the most current, most accurate story. But the system would automatically note if a story had been changed post-publication and enable people to click on a link showing previous iterations, with changes highlighted.
Rosenberg, who runs the error-reporting site MediaBugs, called it the “best compromise of the flexibility that the Web offers and the kind of permanence that you want your reporting to have.”
This intriguing idea could end the memory hole parlor game, but you're unlikely to see it implemented on any news sites. Not due to the technological and design challenges, which are significant (unless your site uses WordPress, for which a plugin already exists). But because our industry isn't ready to tolerate the downside of such radical transparency for the potential benefit.
Besides enabling people to see early coverage of developing stories, article versioning would reveal the many errors that are fixed on the sly without corrections notices. And it would shine a light on all the other tweaks, from subtle shifts in wording to changes that are debatable factual errors.
Rosenberg believes this transparency will bolster media credibility. But it could also strengthen arguments that the media can't keep its stories straight.
A simple comparison between two versions of a news story will show what was changed, but it won't tell people why. Explaining these revisions, some of which editors would consider trivial, would create more work; ignoring readers' questions would alienate them and raise more suspicions.
No wonder news sites aren't falling over one another to give the public another reason to doubt the information they publish every day.
A solution in search of a problem?
I asked editors at a couple of news sites what they thought about the idea. Neither David Plotz, editor of Slate, nor Josh Marshall, editor and publisher of Talking Points Memo, expressed much interest. They share Rosenberg's commitment to accuracy, but they believe they're satisfying it by correcting errors and notifying when a story has been fixed or changed.
At Slate, errors are fixed in the text of the article and a link is added to a correction notice at the end of the story. Post-publication revisions are noted in the text. All corrections are collected in a separate section, and they're emailed once a week to subscribers.
“I think the transparency in correcting mistakes of fact is valuable,” Plotz told me by phone. “The corrections policy we have in place ensures that, and it ensures that with a minimum of inconvenience to readers. … The goal is reader clarity; it is not massaging our own ethics organs.”
Marshall told me via email that he is “a little skeptical” of article versioning, in part because it would highlight post-publication copyediting on Talking Points Memo:
“We do a decent amount of copy-editing post-publication -- not something we're necessarily proud of, but we do. Sort of a byproduct of our publishing speed and focus on fact-checking/story editing over copy editing. We will also edit a malapropism after the fact or a particularly awkward sentence etc. … Perhaps there's part of me that doesn't want it to be totally easy to pull up that we first wrote 'know' and not 'no' or made some other silly copy-editing error.”
“...Obviously if we get something wrong and we need to correct or we add or remove something that materially changes the story, that's another matter entirely. And we have clear internal policies requiring ed[itor] notes telling readers a story has been revised or corrected, etc. And we're pretty rigorous about this.”
Or a secret of online publishing?
Slate wasn't always so diligent. Nor, apparently, was Politico.
Last year, after people criticized Politico for removing a couple of paragraphs from a story (again, the parlor game), a Slate contributor wrote a program to scan Politico.com to see how often stories were changed after publication.
Over three weeks, he counted 12 instances in which stories had been altered significantly without corrections or revision notices. The story caused a bit of dust-up, with Politico posting several corrections and Slate posting a few of its own.
Trouble is, people at Slate were doing the same thing. At the time, Slate allowed errors to be corrected without notification if the error was discovered internally within 24 hours of publication. At the time, Plotz called the hypocrisy “embarrassing,” and he revised the policy.
Yet he does suspect that “a lot of sites are sending things down the memory hole ... that people are making things disappear and they are not notifying readers.” So do I.
You could say that the reason you don't need article versioning is that you already have a corrections policy. But that's one reason you need it – to show skeptical readers that you are following the policy.
In a way, article versioning is the equivalent of Ronald Reagan's oft-used phrase about dealing with the Soviets: “Trust, but verify.” (My apologies for comparing the media to the Soviets.)
The tough part for news sites is that such a system would illustrate changes that lie in the no-man's land between a factual error and the factual ugliness that is in the eye of the beholder.
A versioning system, as Rosenberg notes, could enable editors to add notes explaining such changes. But the day of an editor is filled with five-minute tasks that delay the process of editing and managing content.
A true digital archive
The journalist's version of Donald Rumsfeld's line that “you go to war with the Army you have” is, “you post your story with the information you have, not the information you'll have when the spokesman calls you back with an ironclad account.”
“You should be free to constantly improve your story,” Rosenberg said. “The only reason you wouldn't want to do this is some concern for accountability, which you should be concerned about. And if you have a versioning system in place, you never have to think about that. You're never hiding anything.”
The Web, Rosenberg wrote, “is both up-to-the-minute and timeless — ephemeral and archival.” So are news sites. We focus on the latest updates, but the collective store of news is equally important.
Print newspaper archives basically served one purpose: to provide a history of a community, one news story at a time. Though accessible to the public (at the paper or the public library), the archives were principally used by reporters.
Online archives do other things, too. Previous news stories provide context by being linkable, that day or many days in the future. And because they're easily accessible to the public, they provide accountability of those who make news and those who report it.
When news sites change blog posts and articles and discard the previous versions, they undermine the authority, context and accountability of their online archives.
Deleting material after people have linked to it breaks a small piece of the Web. And it sweeps aside some versions of the news in favor of others. The stories that remain often are more complete and accurate, but sometimes they drop details that were relevant in the past and may prove important in the future.
A versioning system enables each blog post or article to be its own archive, becoming a record of a changing story – whatever reason it changed.
Do mea culpas help or hurt our credibility?
The ultimate question for news organizations is whether revealing the changing nature of their content will increase their credibility or further undermine it.
We know that inaccuracy – from simple errors to major misunderstandings – hurts our credibility. So will we damage it further by revealing all the mistakes we caught without anyone noticing?
Will readers trust us more when they realize how carefully we choose our language, or will they see it as subtle efforts to color the news? Will they be comforted when they see stories with revision notices? Or will it add to their misgivings?
Rosenberg believes that absolute disclosure will bolster credibility, and sites that decide to post correction notices even for borderline cases will be seen as conscientious. I could just as easily see how a such a site could be unfairly accused of having a culture of inaccuracy.
Moreover, Rosenberg says users will become increasingly comfortable with versioning systems and will wonder why Wikipedia is more transparent than their local news site. Users may become that sophisticated in their use of the Web. Unfortunately that doesn't mean they'll have an equally sophisticated understanding of the messy, complicated ways that news is reported.
A versioning tool is a good start, but ultimately questions of credibility will not be solved by technology alone.