BS detection for journalists: Rough Notes from an ONA 2011 Panel
Things to consider when verifying socially mediated information
The following information came from a couple of panels at the Online News Association during September 2011, particularly the panel titled “BS Detection for Digital Journalists.” Panelists included: Craig Silverman of RegretTheError and Mandy Jenkins of the Huffington Post. Some material was also used from remarks made by Andy Carvin at NPR during the conference. (And for a great recent example of all of this playing out, check out the Reuters photo of a rebel firing an RPG, which has been accused of being fake! Good discussion under the photo about why the photo may be real or may have been Photoshopped!)
This is the second in a series of three “rough notes” from the conference.
General Thoughts on Detecting B.S. online
— You may not know every single detail through your social media networks, but you can know a lot. Alas, you have to sort it out. That’s your job.
— Do not blindly retweet without verification or some kind of indication that it is not confirmed.
— Often it’s just as great a story if you can debunk something everyone else is saying is true.
— Be transparent with what you do not know!
— Use social media as another source of verification for your observations. Tweets and status reports can help contextualize what may otherwise be a myopic view of the world when you are in the thick of some event.
–Corroborate the story in as many ways as you can. Check the scanner or police reports for news of the incident. Search twitter.com to see if others are saying the same thing. Crowd source it, making sure to indicate the lack of verification of the story.
— If you make a mistake: fix it and retweet the correction several times. Direct message all of the people who retweeted the original tweet. Fix it everywhere you posted it, and make the correction go as viral as the original tweet. Embarrassing but necessary.
— If you see 20-30 original tweets saying the same thing, it’s probably true. (Thus, one could say that volume adds to the sense of authenticity. And yet still, one must check it out. Just make sure you are not verifying via the original tweets)
— Andy Carvin says he sees his Twitter feed as part of his fact-gathering process. “A lot is not verified and when that happens, I put, ‘source?’ And then we try to hash it out.”
— Observe who your followers are retweeting. Follow those people, too.
— Be wary of people who use journalistic terms by people who are not journalists. I.e. “confirmed.” These terms are often not accurate in the journalistic sense and used to gain attention.
— Before you retweet, check the original poster. Who is it? Check their profile. Check how long they have been around. Newer pages, profiles, sites should be approached with suspicion. Look out for parody accounts with telltale misspellings in the handle.
— To check the longevity of a tweeter: http://howlonghaveyoubeentweeting.com/
— Check on their profile for how frequently they update. You want to determine whether they are organically a part of ecosystem.
— Is there a photo on the profile of a real person?
— How many friends and followers do they have? Who are they following?
— Who else is this account interacting with? If they are replying to people and having a good bit of back and forth with other people, that adds to their credibility.
— Check the Klout score. The higher their score, the more active they are.
— Google their name with keywords like “spam,” “lies,” “bots.”
— What other accounts are associated with them? Are they on Facebook or LinkedIn? Do they have a blog?
–Use sIte identifiers like “hoverme,” an add-on that works with Chrome, Firefox and Internet Explorer.
— Direct message the tweeter and ask for a phone number, and then call them. Ask them what they heard firsthand. Walk them through what they saw from the beginning. Find out who else might have witnessed it. Was there a crowd? Was there someone else with them to whom you could talk? Did they mention why they were there? Consider what they tell you in relation to your gut; does the timeline of events work out? Does the information check with other information you gather about the day/place/site?
— Check out what they said later, after the original tweet.
— Before you retweet or publish the information, evaluate the options: Is this the entire story? Is it really worth the risk of what I am running with if it is not verified? Do not just check one thing: triangulate! Most of time it is not worth rushing it out there. You get a better story by being the person to go through all these steps and a have more solid account.
— Go back to the original source of the twitter chatter. (You do this by searching for the keywords in most of the tweets, such as “fire” or “police” and follow the list chronologically until you see one that has not been retweeted.)
On Verifying YouTube Videos:
— Note the scene and all the details in the photo and then corroborate, corroborate, corroborate: Are people wearing coats and it is summer? Is there a fountain pictured that should not be there? Examine weather reports, language at people speaking. License plates on automobiles, vehicles themselves, do these belong in this place?
— Ask the poster if they have more photos. They generally do not just take one photo of it,
— Check the Exif data, which is information that is embedded on all images.
— Also Error Level Analysis will let you see if it has been edited in Photoshop.
— Check out tineye.com, a reverse image search site that can help determine if photos have been used before elsewhere.
— Call everyone in the relevant neighborhood and send them the picture. Don’t be afraid to ask the resident to go down the street and take a look at wherever the supposed incident had taken place. You can use neighborhood directories to find phone numbers (most newsrooms have a bunch on hand).
Verifying Content on a Website
— Do you remember the story, “Are Internet Explorer users dumb?” This story went viral, though it was erroneous, because it reinforced an already existing attitude. Good hoaxes try to get into the sweet spot of what people already believe. However, one key giveaway that the story was wrong was how striking the findings were; the statistical difference between the IQs of IE users and Firefox was just too large and should have raised a red flag.
— When was the domain registered?
— Is it difficult to find out who owns the site or who is running it?
— Go to the Internet Archive: how long has the site been there? Has this site changed radically recently?
— Check out the page rank by Google: if the page rank is good, the score means the site is linked to by a lot of people and is a high influencer and could be an indication of the site’s authority and authenticity.
— What is the level of interaction? Are there comments? Are people bookmarking it on delicious or diigo? These are all good indications that the site is credible.