Correct, don’t delete, that erroneous tweet

Over the weekend many news organizations reported, erroneously, that Rep. Gabrielle Giffords was dead. These reports don’t seem to have originated on Twitter. But many spread there — and now they’re occasioning a round of head-scratching over how to handle retractions and corrections in this new communications format.

This happens with every new phase of communications-technology evolution. Twitter, with its speed and popularity and intermingling of professional and personal channels, presents some modest new challenges to accuracy practices. But for journalists there should be little confusion about the answers.

At Lost Remote, Steve Safran writes:

We ask: is deleting a tweet after the fact a lack of transparency, especially if any subsequent tweets don’t admit the error? Is a news organization obliged to tweet that it was wrong? Does the retweet function make such actions moot? We strongly believe in transparency, as do many of you. But whether deleting tweets is a responsibility or not, and whether a news organization must tweet that it was wrong, should lead to serious discussions in all newsrooms.

For a private individual using Twitter, it might make sense to delete a message that you later discovered was in error. But for anyone tweeting as part of a professional media job, representing a news organization on Twitter, or using Twitter to do journalism independently, the course here ought to be plain: It’s almost always better to correct than to unpublish. Removing information you’ve already disseminated — sometimes called “scrubbing” — always leaves open the possibility that you’re trying to hide the error or pretend it never happened.

The folks at WBUR have it just right:

We have decided NOT to delete the erroneous tweet, because it serves as part of the narrative of this story. Facts can change fast when news is breaking, and that leads to errors. We need to own the error, not hide from it. But we also need to rectify the error and explain ourselves to people who trust us. Deleting the tweet would do more to harm trust than preserving it would do to harm truth.

According to the chief argument in favor of tweet-deletion, if you leave a bad report lying around your feed, you’re tempting others to retweet it; if you delete, you’ll inhibit this viral repetition of misinformation. That’s a reasonable position. But there are alternatives to simply zapping the bad tweet and scrubbing the record.

The same technologies that force these problems on us can also help us solve them. On the Web, for instance, publishers can use versioning tools to keep a corrected edition of a story front and center while maintaining a trail of accountability. Similarly, Twitter users can use Twitter itself to correct the record while preserving it.

For instance: say your newsroom had sent out a tweet that read, as NPR’s did:

BREAKING: Rep. Giffords (D-AZ), 6 others killed by gunman in Tucson

You could send out a followup message, preserving a record of the error while correcting it:

CORRECTION Giffords wounded, in critical condition RT @NPR BREAKING: Rep. Giffords (D-AZ), 6 others killed by gunman in Tucson

Then you could reasonably go back and delete the original. This might be a useful tactic to curtail the spread of bad info. But it still flattens the record a bit, since the original message’s timestamp (and possibly other contextual data) would vanish.

Better yet is the idea floated by “Being Wrong” author Kathryn Schulz (@wrongologist) in this Poynter interview by Mallary Jean Tenore: “Why not have a ‘correct’ function (like the ‘reply’ and ‘retweet’ functions) that would automatically send a correction to everyone who had retweeted something that contained an error?”

Such a tool would dragoon the engine of viral misinformation back into the service of truth. You can say, “Hey, it’s just Twitter, what’s the big deal,” but experience suggests that arguments about Internet tools that begin with “It’s just” will get disproved over time. Twitter is beginning to mature into the rapid-response nervous system of our news world. It needs and ought to have a function like the one Schulz proposes.

BONUS LINK: Craig Silverman has a valuable roundup of links relating to errors in Tucson shooting coverage.

[I've continued discussing this topic in a followup post.]

Related

Post Revisions:

Get Scott’s weekly Wordyard email

Comments

  1. I disagree that it’s almost always better to correct or delete. I’d actually encourage deleting, as long as corrections acknowledge it.

    Say the newsroom had put out the tweet saying Giffords was killed. Later, a correction goes out, but the original lives. That means there’s a chance some people will….

    1) Never see the correction for some reason
    2) Go directly to the tweet for some reason, say from a retweet

    In both cases, not deleting the original helps spread incorrect information.

    In contrast, let’s say you delete the original tweet but also put out at the same time a correction. That means:

    1) You’re not hiding that you had something wrong, but…
    2) No further people are inadvertently exposed to the incorrect information
    3) You kill the original tweet across the web, if it was retweeted using Twitter’s retweet feature

    I think it’s dangerous for news organizations to err on the side of “we’re not trying to scrub the record” versus getting things right. It’s a similar case with news articles. Rather than correct original incorrect articles — some news organizations will issue a correction that’s not linked to the original article, meaning people may arrive into the original article via search without any context.

    And yes, Twitter should allow for corrections.That’s long overdue. But even then, you’d still need to issue a fresh tweet in order to alert many people.

  2. I”m with Danny on this one… We don’t have to treat new media in quite the same way we would old media (we’re still figuring it out, right?). Deleting AND tweeting a correction does the job. To NOT delete the old tweet could even be considered irresponsible. After all, why would we have the ability to delete in the first place if we weren’t meant to use it.

  3. It’s easy to tweet a correction, but it’s impossible to keep people from retweeting erroneous info.

    A correction function would definitely be a step in the right direction.

  4. What about screenshoting the original tweet in permalink form, then tweeting a correction with a link to the screenshot. A la: “Correction to our previous tweet (http://bit.ly/screenshot_of_tweet). Giffords wounded, etc…”

    Then you could delete the original and thus minimize the viral spreading of the it, via retweets, etc. But you would still have an\ quality record of the original error and wouldn’t be truly scrubbing it.

  5. If the issue is delete or not, why not have a “still valid” flag for tweets. Rather than a hard delete, the tweet could have an indicator that it’s no longer considered valid, possibly with a clarifying URI.

    Rather than deleting and “destroying data”, and without versioning tweets like wiki pages, simply having additional metadata may be able to solve the problem. Of course, it needs client updates to support it, but if it could be done in a backward-compatible way, there’s no need to wait for client support to implement it.

  6. A correction feature that notifies anyone who retweeted would surely be open to abuse outside the hands of news agencies. For spinning something into self promotion or for a ‘second hit’ in marketing. It’s not hard to imagine situations could be set up just to enable an ostensibly legitimate correction of a tweet that serves one of the above purposes.

    Establishment of truth is always going to be a problem in any kind of media, but I think it does present a particular issue in the ‘viral age’. Though how much responsibility is down to the news agency and ensuring truthfulness of stories before they are reported?

  7. Scott Rosenberg

    This is a great thread, thanks to all and esp. to Danny for the careful counterargument. I’nm going to let it all sit in my brainpan a bit longer before responding.

  8. Delete…but post a screenshot of the deleted tweet with a link to the correction, and some wording about the situation. Danny is right—you are responsible for spreading misinformation if you leave misinformation available with no context. And that’s the problem with Tweets—they are often found, viewed and used contextless.

  9. If the argument for deleting, not correcting, is that many people won’t see the correction then unfortunately, with the way Twitter works, deleting won’t help you out much neither.

    Manual/quote retweets won’t be affected by a deletion.

    Deletes can take hours if not days to propagate. Twitter doesn’t even guarantee an “eventually consistent” system. Native retweets of the now deleted tweet also can take a good deal of time to propagate.

    Twitter clients don’t look for deletions. Twitter.com may not show the tweet anymore but other cached clients do not have to update. They don’t get a list of “deleted tweets”.

    Most curation tools, news agencies, blogs etc. copy the tweet text. Deleting the tweet on Twitter does nothing for them.

    Twitter doesn’t have an edit feature not because of feature choice but because of technical efficiency limitations. It is costly to maintain a canonical tweet and update all copies of it as it is. The delete feature is costly enough as it is.

    Even having said all that, I prefer a correction to a delete. It maintains the historical record and illuminates how stories happen.

  10. I favour Alex Byers’s idea: screenshot the incorrect tweet, delete it, and then issue a correction with the accurate info and a link to the screenshot. That preserves data like timestamps, minimises the spread of incorrect info, and is transparent. Can’t see any flaw to this plan.

  11. A screenshot doesn’t preserve much. Timestamp is “6 hours ago” and would require the screenshot being timestamped reliably. You loose any other payload data with the tweet (geo,. It isn’t easily searchable. You lose the context in which the original tweet was made. You lose who retweeted it, who replied to it, who favorited it.

    I realise I am being technical here but if we ever did have a serious case of misinformation then screenshots of deleted tweets aren’t going to cut it. I’d far prefer the flow is preserved and a correction tweeted out.

    BTW does anyone know if the Library of Congress tweet collection takes into account deletions?

  12. I’m a social-justice activist tweeting as @creepcops in order to document (or “curate”) and censure police misconduct and educate people about it. Mistakes can creep (uh) in for a variety of reasons, and they need correcting. I get RT’d a lot.

    As an example, I recently found a story in Camden, NJ. But the newspaper source was in nearby PA. The page I was reading stated only “Camden” with no state. (Online sources are notoriously bad about identifying details of locations for people arriving at the page from all over the world.) I put “Camden, PA” in Google (with quotes) and got lots of hits and some maps. So I wrote that in the tweet’s dateline.

    Of course, I soon saw my mistake. I recreated the tweet with “NJ” in place of “PA” and deleted the first. I really can’t see a better way to do that. People who RT’d the original would have their RT erased when I deleted the first tweet, though.

    I watch in general to see who RTs stuff, but it’s unwieldy to do a manual control even with 1900 followers, and when that grows further it will only be harder. I have, in the past, DM’d a couple of people when I’ve had to perform a similar “tweetectomy” – so they could update their RTs if they chose to.

    I like the idea for new capability in the Twitter API to allow a “correct” function. Indeed, I have always been for allowing edits to tweets. But if they were somehow explicitly marked as having been revised, well that would be just fine. A serious problem with deleting is that later tweets can easily be dependent on earlier ones in a pseudo-threaded discussion.

    I do agree that major breaking-news tweets such as about an assassination attempt, if first wrongly reported, should be retained (and, if at all possible, marked) as acknowledged errors that are later corrected.

Trackbacks

  1. [...] Para Scott Rosenberg, NPR hizo lo correcto. Mantener el tweet original y publicar una corrección que haga referencia a la información errónea. En LostRemote, sin embargo, recogen las razones por las que publicaciones como CNN o PBS optaron por borrar todos los mensajes con información falsa. Un argumento a favor [de borrar] es para detener o ralentizar el reenvío de mensajes. Pero esto es difícil, si no imposible. Y es tentador aunque poco práctico pedir a un equipo de personas que se dediquen exclusivamente a controlar esos tweets. Durante horas después de que se informara que Giffords estaba viva, la gente seguía leyendo el tweet diciendo que estaba muerta, reenviándolo a sus amigos sin ver la actualización. [...]

Post a comment