Tracking Back a Text Message: Collaborative Verification with Checkdesk
Written by: Craig Silverman
During the days of heavy fighting and bombardment in Gaza and Israel in the summer of 2014, this image began to circulate on Twitter and Facebook:
It purported to show a text message that the Israeli Army sent to residents of an area in Gaza, warning them of an imminent attack. Tom Trewinnard, who works for the non-profit Meedan, which is active in the Middle East, saw the image being shared by his contacts.
“This was one image that I saw quite a lot of people sharing,” he says. “These were people who I would expect not to share things that they hadn't checked, or that looked kind of suspicious.”
It seemed suspicious to Trewinnard. A few things raised questions in his mind:
- The message was in English. Would the IDF send an English message to residents of Gaza?
- Language such as “We will destroy your house” seemed too stark, even though Trewinnard said he finds the IDF’s English Twitter account is often very blunt.
- He wondered if someone in Gaza would have “Israel Army” saved as a contact in their phone. That is apparently the case with this person, as evidenced by the contact name in the upper right hand corner.
- The image has a timestamp of 9:56 in the bottom left hand corner. What’s that from?
Trewinnard’s organization is developing Checkdesk, a platform that people and organizations can use to perform collaborative verification. He decided to open a Checkdesk thread to verify the image, and use it to track the verification process for the image in question.
He kicked off the process by sending a tweet from the Checkdesk Twitter account that invited people to help him verify whether this was a real text message:
“The Checkdesk account has less than 300 followers,” Trewinnard said. He didn't expect an onslaught of replies. But a few retweets from people with a lot of followers, including @BrownMoses, inspired others to take action.
The Checkdesk tweet included two specific questions for people to help answer, as well as an invitation for collaboration. Soon, Trewinnard was fielding replies from people who offered their opinion, and, in some cases, useful links.
He was also pointed to an Instagram account that had shared what appeared to be a real message sent by the IDF to someone in Gaza close to two years earlier:
Trewinnard was able to verify that the Instagram user in question was in Gaza at the time, and that Israel was carrying out an operation in Gaza in that timeframe. He also saw that the same image had been used by the credible 972mag blog.
The above image provided a valuable bit of evidence to compare to the image he was working to verify. It differed in that the above message came in Arabic, and showed that the sender was identified by “IDF,” not “Israel Army.” Trewinnard also said the tone of the message, which warned people to stay away from “Hamas elements,” was different than the language used in the message they were trying to verify.
This all suggested the image he was working on was not real. But there was still the question of where it came from, and why it had a time stamp in the bottom corner.
Trewinnard said he tried doing a reverse image search on the picture to see where else it had appeared online. But he didn't immediately click through to all of the links that showed where it had appeared on Facebook. Another Twitter user did, and he found a post that showed conclusively where the image had come from:
The Facebook post includes a video that clearly shows where the text message came from. It was shown in a short film clip that is critical of Israel. The numbers in the bottom left hand corner correspond to a countdown that takes place during the video:
“So there had been these flags ... but this guy found the actual source of the image,” Trewinnard said.
He said that the entire process took roughly an hour from his first tweet to the link to the video that confirmed the source of the image.
With an answer in hand, Trewinnard changed the verification status of the image to “False.”