New York Times public editor Art Brisbane today addresses an issue that MediaBugs and I have been talking about for a year: the need for news organizations to maintain a record of the changes they make to published stories.

I’ve argued that posting such “versions” of every news story — the way Wikipedia and every open source software project does with their own work — would help newsrooms regain public trust and free journalists to update their work more vigorously while staying accountable.

Brisbane seems to agree, but sounds doubtful that the Times is going to do this any time soon.

Right now, tracking changes is not a priority at The Times. As Ms. [Jill] Abramson told me, it’s unrealistic to preserve an “immutable, permanent record of everything we have done.”

I know the Times has tons of claims on its resources. Jill Abramson has a million demands to juggle. But let me respectfully dispute her “unrealistic” judgment.

Versions of stories are just data. For the Times, or any other website, to save them is a matter of (a) storage space and (b) interface tweaks to make the versions accessible. Today, storage is cheap and getting cheaper, and Web interfaces are more flexible than ever.

Really, there’s nothing unrealistic about preserving an “immutable, permanent record” of every post-publication change made to every story.
Wikipedia — a volunteer organization run by a variety of ad hoc institutions — can do it. Any WordPress blog can do it. It seems peculiarly defeatist for our leading newsroom to shrug and say it can’t be done.

By making story versions “not a priority,” the Times is essentially abdicating its longstanding status as our paper of record as it makes the transition from paper to digital. I doubt that’s what its leaders intend to do. The more they ponder this, the more I think they’ll see that a versioning system for news is not only valuable but inevitable.

The journalism industry ships lemons every day. Our newsrooms have a massive quality control problem. According to the best counts we have, more than half of stories contain mistakes — and only three percent of those errors are ever fixed.

Errors small and large litter the mediascape, and each uncorrected error undermines public trust in news organizations. In Pew’s last survey in Sept. 2009, only 29 percent of Americans believed that the press “get the facts right.”

Yet the tools and techniques to fix this problem are known and simple. I’ve been working in this area for the last two years. Here’s a distillation of what I’ve learned: three basic steps any online news organization can take today to tighten quality control, reduce errors and build public trust.

Link generously

A piece without links is like a story without the names of its sources. Every link tells a reader, “I did my research. And you can double-check me.”

• Read more on the value of links: In Defense of Links.

Show your work

The news isn’t static, and online stories don’t have to be, either. Every article or post can and should be improved after it’s published. Stay accountable and transparent by providing a “history” of every version of each story (a la Wikipedia) that lets readers see what’s changed.

• Read a longer argument for the value of versioning.
  Or try out the WordPress plugin.

Help people report your mistakes

The Internet is a powerfully efficient feedback mechanism. Yet many news organizations don’t use it. Put a report-an-error button on every story: It tells readers you want to know when you’ve goofed. Then pay attention to what they tell you.

• Get some report an error buttons at the Report an Error Alliance.
  Or use the MediaBugs widget.

Why aren’t these practices more widely adopted? Here are four reasons:

(1) Workflow and tools: In many newsrooms, especially those still feeding print or broadcast outlets, it’s still way too hard to fix errors or add links to a story for its Web edition. And content-management systems don’t yet offer corrections and history tools “out of the box.”

(2) Denial and avoidance: Other people make errors. Many editors and reporters don’t believe the problem is serious, or think it doesn’t apply to them. And most don’t understand how badly their Web feedback loop is broken.

(3) Fear of readers: Many journalists view readers as adversaries. The customer they feel they’re serving is an abstraction; the specific reader with a complaint is “someone with an agenda” whom they have a duty to ignore.

(4) Where’s the money? Many media companies are in financial free-fall. Correction systems and trust-building tools don’t bring in revenue directly, and they eat up product-development time and money.

These are serious obstacles. But journalists will never regain public trust unless we overcome them.

Ask journalists what sets them apart from everyone else sharing information online and we’ll say: We care about accuracy. We correct our mistakes. In a changing media economy that’s challenging the survival of our profession, we need to follow through on those avowals. Otherwise, we shouldn’t be surprised when Pew’s next biennial survey of public trust in the media shows even more dismal results.

[Crossposted from PBS MediaShift Idea Lab. This edition employs all three techniques I mention.]

[cross-posted from the PBS MediaShift Idea Lab]

A window of opportunity is open right now for online journalists to build accuracy and accountability into the publishing systems we use every day. To understand why this is such a big deal, first hop with me for a minute into the Wayback Machine.

It’s the mid-1990s. Journalists have just arrived on the web. They’re starting sites like Hotwired and Pathfinder, Salon and Slate. They’re doing good work, but also, inevitably, making mistakes. Their customary corrections routine — post a notice in the next edition or issue — makes no sense in the new medium, where stories are just files on servers or data in databases, and fixes can take effect instantly and invisibly.

Editors at the dawn of the web understood they had to be accountable for changes they made to published stories, and so improvised a routine for handling substantive corrections: Fix the problem; place a notice on the story page indicating that you’ve fixed it; and — this step was only taken by extra-conscientious organizations — add a notice to a separate page logging the fact of the correction (and linking to the corrected story).

>p>Fast-forward to the present. The web’s publishing environment is vastly more complex, flexible and elaborate. But when it comes to corrections, virtually every news site still handles things the way we did 15 years ago: Go into the story, often by hand (i.e., by adding to the body of the story text), fix the error, and append a correction notice to the story top or bottom. Then, if your site has a separate corrections-listing page, go into that by hand and add the notice there. Insert any cross-links. Republish the story and the corrections page. And you’re finally done.

The process is cumbersome, to be sure; it’s also not smart. Most publishing systems don’t actually “know” that the story has been corrected. There’s no data stored that distinguishes a corrected story from, say, one that’s been altered in some other way. The typical content-management system software package will track each successive edit or revision to a document, but it doesn’t distinguish garden-variety edits from formal corrections.

For years now, I’ve dreamed of a smarter publishing software tool that would handle corrections intelligently and seamlessly as part of the publishing cycle and editorial workflow, rather than as a clumsy kludge. One goal, certainly, is to make editors’ lives easier. If corrections can be handled with less fuss, maybe news sites will be less reluctant to make them.

But an even more important goal is to give journalists and the public better information about corrections. Once corrections are treated as data, developers can do things with them — say, allow readers to sign up to be notified of corrections for a site, individual story or story category; or create display boxes that automatically link to the half-dozen most recent corrected stories. The ultimate purpose of all this is for news organizations to demonstrate accountability and transparency to a public that views them with sparse and dwindling trust.

So when I read about the new Armstrong CMS project, I got excited. Armstrong is an effort by the software teams at the Bay Citizen and the Texas Tribune to build a new-model, open source publishing system for local news sites. It’s working off the highly regarded Django content-management framework, funded by a $975,000 grant from the Knight Foundation, and building on existing work already in use at the two sites.

The Armstrong project has a chance to create a new standard for corrections for the entire field of web journalism. I asked Brian Kelley, the Bay Citizen CTO who is a co-leader of the project, whether Armstrong had plans for corrections yet. He suggested that, because many organizations have different needs, Armstrong’s open plug-in and extension options might be the best way to handle the corrections process.

Maybe so. At MediaBugs we certainly plan to explore this route with Armstrong as we have with other partners; our MediaBugs widget and WordPress plug-in are already in use on a handful of news sites.

But there’s a bigger opportunity for the Armstrong community here: They can build a smart correction-handling process into the heart of the tool they’re creating. The best practices in this area are widely understood and agreed upon; why not bake them into the technology? No one, to my knowledge, has done this before in a free, open source publishing system. (If there are proprietary systems that do a better job, I’d love to hear about them.)

Here are the basic features I’d want any corrections tool to provide:

• Editors should be able to correct published stories by checking a box or clicking a button on an edit screen. If the system has a permissions hierarchy, then managers should be able to enable or disallow the option of making a correction.
• Editors who are correcting a story are taken to a screen or overlay that lets them enter the text of a correction notice. The software would automatically record the date and time the correction was made.
• Once the correction notice is entered, the editor is prompted to make whatever edits are required in the story text itself, and to save them. Editors would then have to republish the story, following whatever their site’s routine might be.
• Ideally, a corrections system like this is part of a larger scheme for tracking and presenting all post-publication changes to each story. The database would record the changes made to a story as part of the correction process in a special way — that is, it would know that this particular revision is not just any old change but a formal correction.
• Site designers and managers have the option of building a self-updating corrections page that automatically pulls in corrections notices and links back to the corrected stories.

That’s it! None of this is particularly challenging as programming or design work. My experience is that when I describe what’s needed to most developers, they’re not interested — the problem’s too “trivial.” Maybe it is — but not to the editors I’ve talked to, who groan about the pain their software inflicts on them whenever they try to do a correction the right way.

Each time we rewrite the software used to publish news on the web we have another chance to raise the bar for the whole field. I’m crossing my fingers that Armstrong will be the project to make smart corrections a reality.

Last week I was glad to see Arthur Brisbane, the New York Times’ public editor, call for the paper to post all its policies and standards for news out on its website where everyone can see them. I hope your first thought hearing this matches mine: “You mean they don’t already?”

Brisbane’s column only reminds us how stubbornly most news organizations still cling to their opacity. The Times is relatively forward-thinking about corrections policies, but for it, as for most of our news institutions, public accountability remains more of an aspiration than a way of life.

I agree with Jay Rosen that Brisbane’s suggestion, if implemented, would mark a “major step forward in transparency.” Still, my cheers for Brisbane’s column were muted a bit by its clumsy framing of the challenges the Times and its peers face as they set out to open themselves to public scrutiny.

Brisbane seems to be under the impression that posting a set of policies presents the Times with difficult technical challenges. In reality, the technical and design issues here are nonexistent — posting a bunch of static web pages shouldn’t take the Times staff more than a few days, if not hours. The problems, rather, are organizational: Most institutions are reluctant to expose their inner workings to sunlight. Newsrooms are no different — but they should be. They exist to inform the public and hold powerful people and institutions accountable. They should set an example for those they cover, not just to avoid hypocrisy but also to earn back some of the public trust they’ve lost.

Here is the heart of Brisbane’s proposal:

The Times should step out ahead of its industry peers by creating a reader-friendly portal to its policies on ethics, style and usage, blogging, anonymous sources, social networking and other subjects that readers and journalists care about. I envision a link on the left side of the NYTimes.com home page that would take you to a Journalism Policies page where you could locate topics using a search tool.

A link! A page! Topics that are searchable! There isn’t a content management system in use today that doesn’t make providing such material a snap. The Times probably employs several. Yet Brisbane goes on to assert that “building the portal would require considerable programming time.” Really, it should take zero programming time, a tiny bit of a designer’s time, and a modest amount of editorial time to prepare and organizes the policy pages. That’s it.

The real cost to the organization would lie in the long meetings where editors would have to hash out whether they can really commit, in public, to every avowal of each existing policy. Once you publish detailed policies, as Brisbane points out, you face inevitable “headaches” as the online public begins to compare the paper’s stated policies with its daily practices.

Well, get out the Tylenol. These are the very same headaches that good journalists visit every day on public officials, businesspeople, and everyone else they write about. Goose, meet gander.

By all means, let the Times and its competitors follow Brisbane’s suggestion. But his “policy portal” is a bare minimum, a catch-up-to-the-present move. It’s a small down-payment on the kind of real transparency that we have every right to expect newsrooms to epitomize. News providers should go way beyond spelling out their policies for the public; they should unveil as much of their actual practices and processes as they reasonably can.

One of the tenets of MediaBugs is that there’s value in providing a public, permanent space for discussion and debate about potential errors in news coverage — illuminating a process that has traditionally taken place in the dark. As Brisbane says, this sort of thing provides dividends in trust that journalists today desperately need. It also actively improves the coverage journalists can provide, which should be more than enough reason to do it.

In his latest Sunday column, New York Times editor Bill Keller tries to lay out the Times’ ideals — as distinct from the work of “guerrilla” newsies like Julian Assange and James O’Keefe. Keller’s credo: Verification beats assertion! Correct errors quickly and forthrightly! Who’d argue?

Anyone can embrace these principles; the devil’s in applying them. Of our major news institutions, the Times leads the pack today when it comes to correcting its goofs. It is, I think, the last of our media outlets to accept that the burden of “paper of record” authority means an endless parade of corrections.

But the juxtaposition this weekend of Keller’s self-defense with a particularly glaring Times misstep leaves me with some unsettling questions.

Writing on his Shanghai Scrap blog last week, China-based journalist Adam Minter took apart the Times’ Monday story about electronic censorship in China. The story led with a funny anecdote about cellphone calls in China getting cut off by government censors whenever they utter the word “protest” — even if they’re quoting Shakespeare’s “lady doth protest too much.” The rest of the piece wasn’t about cell-phone monitoring at all, but rather describes a recent tightening of Chinese Internet censorship.

Minter thought the “protest” thing sounded fishy, so he performed an impromptu field-test. He was unable to duplicate the censors’ call cutoff, using three different phrases including “protest”, uttered in succession twice during cell calls to five different recipients in China.

That’s just “anecdata,” sure. But so was the Times’ tale — and if it doesn’t pass this basic sniff test, it shouldn’t be in the paper.

But the story gets messier. On Thursday Minter found a comment on his blog from Jonathan Ansfield, a Times contributor in the Beijing bureau who was listed as one of the story’s contributors. “For the record,” Ansfield wrote, “the contributing reporter’s own tests comport with yours. regrettably his input on the story made little difference.”

Whoa! This doesn’t sound good.

By this morning, the Times had appended an Editor’s Note to the story, explaining that it had failed to mention that the dropped-call anecdotes happened at the Times’ own bureau:

The article did not point out that in both cases, the recipients of the calls were in the Beijing bureau of The New York Times. Because scrutiny of press communications could easily be higher than for those of the public at large, the calls could not be assumed to represent a broader trend; therefore, those examples should not have been given such prominence in the article.

So the Times gets one cheer for dealing with this at all, and dealing with it quickly. But, given the evident breakdown in the editorial process, it also leaves us with a bunch of further questions. We still don’t know exactly what went awry here: How did a solid and important account of Internet censorship get saddled with a misconceived intro about cellphone surveillance — one that not only “should not have been given such prominence” but probably shouldn’t have been published at all? How did the Times’ editorial process override the evident objections of its reporter on the scene?

This is where defenders of the traditional newsroom circle-the-wagons practice pipe up in protest: “What do you guys want? We can’t do our work under a spotlight! Should every editorial argument be aired in public?”

And this is also where reasonable advocates of transparency respond, “Of course not. Not when the system works fine. But when there’s a problem, you owe it to your readers to tell them the whole story of what happened, just as your own reporters would try to tell the whole story of what happened in any other institution that erred.”

One problem is that our newsroom culture still drapes errors in shame instead of handling them as inevitable byproducts of an imperfect business. Keller’s column talks about “taking yourself to the woodshed,” which is, if I recall, the place where you get spanked, or worse.

The Times’ opaque Editor’s Note non-explanation is a symptom of a kind of defensiveness that infects most of our news institutions. Yes, we will correct our errors, say our editors. But first, they insist, prove to us we were wrong! Then, and only then, we’ll grudgingly admit it while doing our best to minimize it. But don’t expect us to tell you the whole story of the process that led to the error — unless it was so scandalous (see: Jayson Blair; Iraq WMD) that we feel we have no choice.

This defensiveness is inevitable; it comes with our humanity. That’s why journalists committed to verification and accuracy need to bend over backwards to counter it.

Keller makes this argument himself in his effort to explain why he considers the “impartiality” of Times reporters to be such an immovable principle. He writes, “Once you proclaim an opinion, you may feel an urge to defend it, and that creates a temptation to overlook inconvenient facts when you should be searching them out.” This is certainly true — and it is why the best opinion columnists make a point of seeking out the the most inconvenient facts and the strongest opposing arguments.

What Keller doesn’t seem to see is that the logic he applies to opinion also holds for fact. Once a news organization proclaims a version of reality, its first instinct will always be to defend it. Trouble is, the defensiveness doesn’t protect the newsroom at all; it actually further undermines the public’s already shaky trust in the journalist’s work. The reader thinks: Why won’t they just tell us what happened? What are they trying to hide?

So now Bill Keller is writing a regular column, and he’s given us his journalistic credo of verification, impartiality and the “business of witness.” Wouldn’t it be great for him to apply those ideals in his own writing about the Times itself? What if Keller used his column to give us forthright, open explanations of how the Times runs off the rails in cases like this Beijing phone-call affair? In other words, not just an editor’s note — an editor’s story.

BONUS LINK: Felix Salmon takes Keller to the, uh, woodshed.

UPDATE: I missed the simple factual error in Keller’s original column, (James O’Keefe didn’t impersonate a Muslim NPR would-be donor, his confederates did). But John McQuaid caught it and filed an error report at MediaBugs. The Times corrected it Sunday afternoon. The paper has let stand a broader misrepresentation Keller made about the O’Keefe affair (NPR exec Shiller, it turns out, offered a derisive description of the GOP via a quote from disaffected Republicans — he wasn’t expressing his own opinion).

Beginning Monday, every new staff-written article on the Washington Post’s website came with a prominent link labeled “YOUR FEEDBACK: Corrections, suggestions?” One click takes the reader to a form for reporting errors or providing other feedback to the newsroom.

This makes the Post the first major U.S. news outlet to heed the call that MediaBugs, Craig Silverman and I made with the Report an Error Alliance, urging news sites to make this sort of link a standard feature, like the now-ubiquitous “share” and “print” links.

Actually, Post managing editor Raju Narisetti explained in an email that the new corrections link has been long planned as part of a broader content-management system upgrade. Conversations about corrections practices at NewsFoo, a digital news conference organized by O’Reilly in December (both Narisetti and I were there, along with Greg Linch, who recently joined the Post as a web producer), triggered more internal discussions at the Post.

When the software upgrade’s January launch got delayed, the Post decided to move forward with a pilot of the report-an-error feature sooner, using a Google Docs form to collect readers’ input. As Josh Young pointed out, this sort of nimble, “grab whatever tool’s handy” web development is typical at startups but less common at large media companies.

The newspaper recently took heat from departing ombudsman Andrew Alexander, who wrote that the paper had “become riddled with typos, grammatical mistakes and intolerable ‘small’ factual errors that erode credibility.”

In the Post’s blog post announcing the new feature, Narisetti said, “It addresses a chronic complaint that we don’t make it easy for our online audiences to engage with us on stories, whether it is about factual issues or other ways to get us to meet their needs.”

The Post is hardly the only major news outlet to hear this complaint. Our MediaBugs survey of correction practices across U.S. news sites chronicled a widespread pattern of obscurity and inaccessibility in this realm.

But don’t news sites have comments? And can’t readers just post there about errors?

In theory, yes. But in practice, on most news sites, the freewheeling debates and endless digressions of comment forums provide an inefficient channel for the public to get reporters’ and editors’ attention about mistakes and problems in stories. The urgent signal that “you got something wrong” gets buried in the noise.

A dedicated channel for corrections reports and substantive complaints can be a labor-saving device for newsroom managers — a means to solicit priceless intelligence from the readers who, collectively, as Dan Gillmor famously says, know more than any individual journalist does.

I asked Narisetti about reactions inside the Post and from the public.

“The newsroom response so far has been good,” Narisetti replied, “in the sense it helps streamline what was an ad hoc process online even as we have very evolved policies on this in print.” After one day, he said, the paper had received “about six” submissions; two were about points of fact, and one has already led to a correction.

“It’s early so premature to judge,” Narisetti added, “but the real goal is to make it easy for our audiences to engage with us.”

That’s a goal worth setting and working on. What media executives call “engagement” is closely related to what other companies call “customer service.” Whatever you call it, journalists aren’t always comfortable with it, but newsrooms desperately need more of it, and the Post deserves hearty applause for pursuing it.

Here are some suggestions for the Post to consider as it reviews this project and its practices evolve:

Use an icon. The Post places its “YOUR FEEDBACK” link fairly prominently, in a right-column inset. But it’s all text, and the reader’s fast-scanning eye doesn’t always locate it on first pass. (Several people who read an early mention of the feature on Twitter mentioned that they couldn’t find it on the page.) Some sort of icon or image would really help. The Report an Error Alliance has proposed one icon as a standard (it’s right at the bottom of this post!), and there’s an advantage to providing an image that users can recognize across many sites (like the RSS-feed icon). But we also know that most sites are picky about the look and feel of icons, and really, any icon is better than none.

Make the form even easier to use. For instance, right now the form requires the user to input the web address (URL) of the story page by hand. In the final version of the feature, I hope the Post will automatically fill in the URL when the reader clicks on the feedback link from a story. Each step that you can streamline for the user is worth taking.

Coordinate the new feedback loop with the old one. The Post site, like many newspaper sites, still displays a kind of split personality between the corrections policies and practices on the print side and those of the online newsroom. For instance, the “corrections” link at the bottom of every Post page points to a list of recent corrections and a block of instructions that appear to relate mostly to the print product. This gets confusing to readers, who don’t understand the organizational divisions behind such disconnections — and shouldn’t have to.

Make the whole process public. The ultimate purpose for a newsroom to open an error-reporting channel is to restore public trust in the process of verification underlying the news report. To earn maximum trust, the channel ought to be transparent: It should be clear to the public whether the news organization is responding appropriately to reasonable feedback. Placing those responses out in the open, in turn, can help defend the newsroom when it becomes the target of unfair or irresponsible critics. Such transparency creates a kind of bedrock of trust, and it’s one of the motivating principles behind MediaBugs. Whether a news organization partners with a neutral organization like ours or prefers to handle the process by itself, conducting the exchange openly keeps everyone more honest.

We know this all requires some new thinking, and maybe even a leap of faith, for many editors. It isn’t going to happen overnight. But we’re convinced it’s the future.

This post originally appeared on the PBS MediaShift IdeaLab blog.

Yesterday’s “Don’t delete that tweet” post occasioned a great debate in the comments. (Go read it now if you haven’t, then come back for my thoughts.)

There are valid cases on both sides of this issue. It seems to me that how you come down depends on the relative weight you choose to place on (a) the short-term benefit of restricting the repetition of erroneous information vs. (b) the long-term benefit of preserving the integrity of a historical record and the accountability of a news source.

Danny Sullivan’s argument for deletion is sensible. The screenshot tactic is intriguing but, as Paul Watson points out, a screenshot is a poor substitute for the original data in context. Given the current state of Twitter technology and tools, I wouldn’t fault any news provider for deciding to delete an erroneous tweet, provided some good-faith effort was made to admit the error rather than hide it.

But — as someone who immersed himself for several years in the history of blogging — I can’t help viewing this subject in the longer context of the evolution of Web media. And the pattern here is hard to miss.

Every new style of online participation is born dangling from a “just.” It’s “just” a tweet, so why bother worrying about deleting it? But every wave of Internet-based communication that preceded Twitter arrived on the scene with a similar sense that it was more ephemeral than what preceded it. Save your e-mail? Why bother? Hey, edit your Web page at will — it’s just data on a server!

Each time, we gradually discover that what we thought was casual has become an essential part of the record of our time. And each time we scramble, belatedly, to retrofit some responsibility onto our practices. Maybe this time we can at least shorten that cycle.

Public tweets play an increasingly important role in our news ecosystem. They tell stories and are part of the story, too. We should minimize tampering with them. We need better tools that might let us correct them responsibly, whether this takes the form of fixes auto-propagating to retweeters or correction notices or revision tracking or all of the above.

In the meantime, we’ll all need to keep improvising. As we do, I hope we’ll all think twice before deleting.

VALUABLE CONTEXT: NPR Ombudsman Alicia Shepard has a great column documenting how the incorrect reports of Gabrielle Giffords’ death started. (Hint: It wasn’t Twitter.)