22 Comments

Summary:

Thanks to incidents like the revolution in Tunisia and the recent shooting of congresswoman Gabrielle Giffords in Arizona, most people have come to grips with the fact that Twitter is effectively a real-time news network. But what happens when that real-time news network is spreading mis-information?

2697110891_30b94c17d3_z

By now, thanks to incidents like the earthquake in Haiti, the recent revolution in Tunisia and the shooting of congresswoman Gabrielle Giffords in Arizona, most people are coming to grips with the fact that Twitter is effectively a real-time news network — like a version of CNN that is powered by hundreds of thousands of users around the world. But what happens when that news network is spreading mis-information? That happened during the Giffords shooting, when the congresswoman was initially reported to be dead, and there are other more recent cases as well: on Wednesday, for example, reports of a shooting in Oxford Circus in London, England swept through the Twitter-sphere but turned out to be a mistake.

The British incident appears to have been caused by two coincidental events: according to several reports, one was an email about a police training exercise involving a shooting in Oxford Circus, which somehow got into the wrong hands and was posted as though it was the real thing. Meanwhile, another Twitter user posted an unrelated message about a TV commercial “shooting” in the area, and the combination of those two things helped to fan the flames of hysteria for a number of hours about buildings being locked down and police sharpshooters being brought in, etc. — which can be seen in the chronicle of tweets collected by one Twitter observer at the site Exquisite Tweets.

In the case of Rep. Giffords, in the minutes following the initial reports of the shooting, a number of outlets reported that the congresswoman had been killed, and these reports made their way onto Twitter — in some cases because the reporters for those news outlets posted them, and in other cases because users heard or saw the reports and then tweeted about them. For hours after the shooting these erroneous reports continued to circulate, even after the reporters and media outlets themselves had posted corrections. Andy Carvin of National Public Radio, for example, spent a considerable amount of time correcting people about the report that he posted, but it continued to be re-tweeted.

This led to a discussion by a number of journalists in the days that followed (including me, in a Twitter chat for web journalists) about how to handle an incorrect tweet. Should it be deleted, to prevent the error from being circulated any further? A number of reporters and bloggers said that it should — but others, such as Salon founder Scott Rosenberg and Carvin (who described his thoughts in this comment at Lost Remote), argued that the error should be allowed to remain, but that whoever posted it should do their best to update Twitter with the correct information. Craig Silverman of Regret The Error, who wrote a post cataloguing the erroneous reports, has also described a way in which Twitter could implement a correction function, by tying any correction to the original tweet.

The problem with this approach, of course, is that Twitter is by definition a stream of content. It never stops flowing, and during breaking news events it flows so quickly it’s almost impossible to filter it all, or be sure of what is correct and what isn’t. And because it is an asynchronous experience — meaning people step away from it and then come back repeatedly — there is no way to guarantee that everyone is going to see an update or a correction, or to stop them from re-tweeting incorrect information.

It’s possible that Twitter might be able to either embed corrections or tie errors and updates together using its so-called Annotations feature, which the company was working on last year and had originally hoped to launch in the fall. But work on that project was apparently put on hold while the company launched a revamped website version of the service and sorted out some other matters. It’s not clear whether Annotations will be revived, but the idea behind it was that information about a tweet — or “meta data” such as location or a number of other variables — could be attached to it as it travelled through the network, something that might work for corrections as well (as noted in a comment below, the Poynter Institute is also working with programmer Adrian Holovaty, founder of EveryBlock, to try and develop a correction mechanism outside of Twitter).

Twitter problem isn’t a new one. Traditional media have struggled with the issue as well, with newspapers often running corrections days or weeks after a mistake was made, with no real indication of what the actual error was. In a sense, Twitter is like a real-time, distributed version of a news-wire service such as Reuters or Associated Press; when those services post something that is wrong, they simply send out an update to their customers, and hope that no one has published it in the paper or online yet.

Twitter’s great strength is that it allows anyone to publish, and re-publish, information instantly, and distribute that information to thousands of people within minutes. But when a mistake gets distributed, there’s no single source that can send out a correction. That’s the double-edged sword such a network represents. Perhaps — since we all make up this real-time news network — it’s incumbent on all of us to do the correcting, even if it’s just by re-tweeting corrections and updates as eagerly as we re-tweeted the original.

Related GigaOM Pro content (sub req’d):

Post and thumbnail courtesy of Flickr user Florian Boyd

You’re subscribed! If you like, you can update your settings

  1. >it’s possible that Twitter might be able to either embed corrections or tie errors and updates together using its so-called Annotations feature which the company was working on last year and had originally hoped to launch in the fall.

    Hmmm I’m thinking perhaps a 3rd party player with an API, rules engine, curation ( NLP and semantics), sentiment of tweets and entity extraction, some type of Filtered Stream Definition Language can beat them to the punch ;)

  2. One value that professional news organizations can add is fact-checks on hot topics — and the news reported by others. Reporters who watch real-time analytics can do this proactively. I would personally consider this good journalism.

  3. This reminds me of the case of EasyDNS a few weeks back. I had a back-and-forth with Jillian York about her tweeting the wrong name – Jillian said “I was typing too fast, sorry!” – I also noticed you (Mathew) did also have an inaccuracy in your post which you corrected afterwards (as seen on the strikethrough) on the same case. Good for you for correcting it.

    However, I’m sorry, “I was typing too fast” is not an appropriate excuse for someone who is trained in journalism, and who practices it!

    The empirical evidence (anecdotal as it may be) that I have gathered is that tweets that correct something will spread at a lower rate than the wrong tweet. That was the case with the reporter who incorrectly reported that Giffords had passed, that was the case with EasyDNS. And whoever made a mistake is often embarrassed so they’re not willing to tweet five, six, ten, 20 times correcting their previous mistake.

    How do we fix this? Think before you tweet/post, that’s how. As someone told me that the Associated Press motto was: “be first, but first, be right”

    1. Thanks for the comment, Raul. I agree that thinking first is always a good strategy — not just for journalists but for everyone. I have been guilty of typing too fast as well, along with other human failings :-) But in the case of NPR and reporting the death of Rep. Giffords, there were numerous reports from a number of different sources that she had died, so that one wasn’t a case of not checking — it was just that those sources turned out to be wrong. It happens all the time, even to traditional media like CNN.

      1. I beg to differ Mathew. What I found from direct experience was that “multiple sources” to journalists turn out to be: twitter and other online source who themselves got it wrong.

        In our case the “multiple sources” included: you, Gawker, The New York Times, The Guardian UK, The Financial Times, Huffington Post and about a million monkeys hitting the retweet button. Did all of these mainstream and prominent internet websites source it via “multiple sources” ? Impossible, since it was completely wrong info to begin with.

        In the Giffords case I recall getting a CNN breaking news alert on the death of the Congresswomen which cited “a source has told CNN”, followed by a sheepish follow-up alert along the lines of, no, not dead but “some reports earlier said she was dead”. So Is “a source” who told CNN she was dead “multiple sources”?

        What did they mean by “some earlier reports” were wrong? They meant the 10 or 20 million email alerts sent out from CNN are “some reports”?

        The funny thing about the Wikileaks episode is people started spreading bad info who had no idea, no concept or any conception of what exactly a nameserver is, what happens if you cut off somebody’s DNS or how to check who does DNS for a domain (whois lookup anybody?) who were all too anxious to retweet what they “knew”.

        So here’s a tip: if you don’t know enough about a given piece of info to verify it or understand it on your own, you have NO business retweeting it.

    2. Raul,

      I would like to correct some “facts” you’ve just written about me here. First, I’m not a journalist (I blog and write op-eds, I rarely report). Second, I’m not trained in journalism.

      I agree with you, of course, but I don’t think you should hold me to a higher standard.

      -Jillian

    3. Raul: ‘the Associated Press motto was: “be first, but first, be right”’

      What if you have to choose one? I hope that the AP would choose right. There are many on social media, including Twitter, who would choose first.

  4. Caveat tweet-reader. If you know what Twitter is, you’ll be aware that some of the content will be false, and that false content can be RT’d. If you don’t know what Twitter is… well, why would you believe something you read on something you don’t know about?

    1. Elizabeth Wellburn Andrew Thursday, January 20, 2011

      I agree with you Andrew. It’s the argument I’ve used with teachers regarding wikipedia. Understand what it is and then use it (and teach kids to use it) accordingly.

  5. “Trust, but verify.”

    And don’t just re-tweet any tweet you see, any story you hear. That should fix it. You know. Like in the good old days of journalism.

  6. Mathew, great post. Since it’s unclear whether Twitter will address this challenge (or should), we have enlisted uber-programmer Adrian Holovaty to collaborate with Poynter & anyone who’d like to build a tool that would flag incorrect tweets and alert folks to an updated/corrected version. We hope people will add ideas to this related Quora question and/or join a chat with Adrian in a few weeks to settle on features:
    http://www.quora.com/How-might-a-Twitter-correction-tool-work.
    Hope you’ll add your ideas,
    Julie Moos (Director of Poynter Online)

    1. Thanks, Julie — and thanks for posting that. I meant to include the Poynter effort and just forgot. I will be interested to see what comes of it — I am a big fan of Adrian and his work with Everyblock and Django.

  7. Newspapers and TV also make mistakes. By being much slower to react than Twitter (12 to 24 hours instead of minutes), the “old” news media tend to make fewer mistakes — but also take much longer to correct them.

    Moreover, the source of an error often “forgets” to correct it. Now, twitter addicts get their news from hundreds of sources at the same time, whereas newspaper readers and TV watchers tend to use only one or two. Thus the latter are more likely to be *permanently* misled than the former, I woudl say.

  8. Great article. It illustrates why we should refrain from the term “citizen journalism” because there is no editorial backstop. In this sense, I disagree with the line above that says that Twitter is “like a real-time, distributed version of a news-wire service.” The AP and Reuters have editorial desks. There is accountability.

    I prefer the term “citizen broadcasting.”

    I wrote about the pros and cons of citizen broadcasting on reputation management a while back. Here is the post if your readers are interested: http://www.jamesjdonnelly.com/2010/09/pros-and-cons-of-%e2%80%9ccitizen-broadcasting%e2%80%9d/

  9. Is it possible to collaborate Tweeter RSS feed into feedburner? http://goo.gl/fb/b4vWN

  10. Ferdinand Burfopolis Thursday, January 20, 2011

    What happens when Twitter is wrong?

    Well, what happens when a friend tells you something that is incorrect?

    Life moves on, that’s what!

    You know how they say :”There are no stupid questions?”

    This was a stupid premise for a “news” story AND a stupid question!

Comments have been disabled for this post