Telegram just dropped on SMAT!
Overview, interesting details, and examples of our new (primarily conspiracy and hate) Telegram data.
Note: most of the graphics in this article come from our back-end databases and so look different than in SMAT-app itself.
SMAT is making available a large collection of primarily far/alt-right, “terrorgram”, and Qanon Telegram channels through both or front-end tools and our public API. This will enable researchers, activists, and journalists to analyze and expose one of the most active networks for hate and mis/disinformation on the internet.
At the time of writing we have ~ 8.8 million posts from Telegram in our system from over 1000 channels but it is also regularly updating. We are also adding new channels to account for a wave of deletions after the J6 coup attempt and efforts by terrorgram and Qanon channels to just rename and continue as before. As you can see from the graph below we have more data the closer we get to the present, but have data stretching all the way back to 2015.
The following are the channels with the most posts in our database. As you can see most of them are automated or large brand accounts such as hateful conspiracy outlets such Breitbart and Zerohedge or repost bots from 4chan /pol/. But we also begin to see German Qanon groups as well.
Y-axis is channel names. X-axis is the number of posts
Among the most active users, we see either primarily or exclusively Telegram bots as would be expected. Among them, we see the Stormfront (white supremacist) and Syria conspiracy connected Al-Masdar alongside Zerohedge again. The user Eugen Müller, for whom we have around 30k posts, is part of our fairly large collection of German far-right and Qanon channels. Their channel’s information alludes to German ultra-nationalist history and identifies as “Patriotic Prussian Elites.” Bellum Acta is a similarly traditionalist and openly antisemitic channel reportedly against the “new world order.”
Y-axis is the username. X-axis is the number of posts.
Another interesting way to aggregate the data is by view count. In the graphic below we show which channels have an individual post with the highest view count in our databases. The highest one is by the owner of Telegram who’s channel bio currently states, “Since my last post, the already massive influx of new users to Telegram has only accelerated. We may be witnessing the largest digital migration in human history. Following this global phenomenon, two presidents started their Telegram channels: The President of Brazil – @jairbolsonarobrasil The President of Turkey – @RTErdogan.” Interestingly the Parler Lifeboat channel, made after Parler was deplatformed, has a post that has ~6.3 million views. Among the most viewed posts are also in the ‘Donald J Trump’ channel and another Qanon channel called “Qlobal-Change.”
Y-axis is the channel name. X-axis is highest number of views for a single post. Parler Lifeboat is highlighted.
The above analyses are a cursory overview of the data and some ways to investigate it but they do not show the full breadth of the dataset. Indeed, many of the most hardcore terrorgram channels are either new, recently defunct, or smaller in size making them not appear in these high-level overviews. However when we do more targeted searches such as “RWDS” meaning “right-wing death squads” we can see some of these including channels mentioned before such as “Gypsy Crusader.” Notably, the “Proud Girls” channel is also posting about RWDS.
Y-axis is channel names. X-axis is the number of mentions of the term “RWDS.”
Another example of a more niche query has to do with P2P technology. The user Ali from the graph about the highest view count is seemingly a moderator for CheckMateNews which is a conservative conspiracy outlet. They are rebuilding much of their media stack into P2P technologies such as Lbry and Odysee. In searching for “lbry” in the data the following channels show up most often and appear to be mostly Qanon-related groups including several German channels, with CheckMateNews at the top of the list and the Qanon “RABBIT RESEARCH” right beneath.
Y-axis is the channel name. X-axis is the number of mentions of the term “lbry.”
Just searching “p2p” more generally reveals a slightly more white-supremacist leaning with channels such as “Bellum Acta” and “Ｂｒｅａｄ Ｐｉｌｌｅｄ (yes breadpilled is a fascist channel trust us)” alongside right-wing conspiracy channels like Zerohedge. This could naively suggest that the Qanon channels are more focused on the specific P2P technology lbry while the white supremacist accounts have an interest in P2P more generally.
X-axis is the channel name. Y-axis is the number of mentions of the term “p2p.”
Examples from the front-end
While the data itself offers interesting outlets for discovery, there is a lot that users can do simply with our free point-and-click front-end tools found on smat-app.com. We have tweeted some examples of this in this tweet thread here (follow us on Twitter!!!):
The graphic below shows a more fine-grained analysis using smat-app’s link counter that searches for the base URLs most shared accompanying the word “kill.” Additionally, users can employ the Activity tool to see which Telegram users are most using a given phrase. Aside from its obvious usage, the Timeline tool can also be useful for things like tracking attempts to doxx or group harassment campaigns.
Advice for using the API
If you want to use our public API to wrangle your own data please refer first to our post on how to use the API here. If you’re contacting us from media or newsrooms please take a look first at our post about media.
The short version of the story is that our interactive API documentation will help you craft simple or advanced queries to look at raw JSON data of our Telegram feed. Everything is nested in “hits” and “_source” for all of our data. As you can see but the key fields for Telegram for most use cases are:
views (note: views count is dependent on when it was crawled)
If you need more data or more advanced interfaces for analysis please reach out to us at firstname.lastname@example.org
As always, you can support our efforts to provide open access to important data services and tools through our Open Collective page.
Thanks for your interest and attention!