If you've been reading the analysis threads @ZellaQuixote and I post, you've probably seen the tweet schedule plots we include. Here's a simple web application you can use to make those plots yourself for accounts of your own choosing. makeadverbsgreatagain.org/allegedly
@ZellaQuixote The plots include 6-7 weeks of history ending with the most recent tweet, limited to the most recent 3200 tweets. If 3200 tweets don't go back the full 6-7 weeks, part of the plot will be blank. The circles are color-coded based on the app used to post the tweets.
@ZellaQuixote Update: a shaded "no data available" box will now be shown if the tweets available via the API (generally the most recent ~3200) do not go back a full 6 weeks. The goal is to prevent folks from mistaking the time range prior to the available tweets for a gap in activity.
@ZellaQuixote The percentage of tweets posted with each app or service has been added to the legend. Additionally, the overall percentage of tweets posted via automation apps/services is shown, along with a red check mark if over 90% (the threshold we've been using for "bot").
@ZellaQuixote Update: we added a test for 24/7 activity to makeadverbsgreatagain.org/allegedly. The score is the percentage of the displayed time range where the account tweets at least 24 hours straight with no gaps of 2 hours or longer. Accounts with scores of 50+% are flagged as automated (red check).
@ZellaQuixote Second update: a "Color Scheme" selector has been added. Two options:
Tweet source: tweets are colored by app/service used (this is the original behavior)
Tweet type: tweets are colored based on one of five categories (standalone tweet, retweet, quote tweet, reply, or thread)
@ZellaQuixote As we evolve this tool further, it's worth considering how to adjust the visualization parameters to make it harder for bad actors to use it to spread disinformation. These plots don't actually show suspicious activity, but the default options are being used to spin it that way.
@ZellaQuixote We've modified the site to automatically determine the appropriate scale for best visualizing the tweet volume of the account being analyzed (you can still manually select a different option if desired). Many thanks to @BeegorBucleor for inspiring this enhancement.
@ZellaQuixote@BeegorBucleor Update the fourth (we think?): we've added detection of repeated tweets to the site. A repetitiveness (is that a word?) score has been added to the legend, and a table of the tweets the account has most frequently repeated within their last 3200 is shown beneath the plot.
@ZellaQuixote@BeegorBucleor The repetition detection only applies to text content of tweets - links, images, video, etc are omitted, as are retweets. The repetitiveness score is calculated as follows:
@ZellaQuixote@BeegorBucleor Due to poor planning, we somehow failed to include @SeanSpammity in the example screenshots of the new "repeated tweets" feature. This has now been rectified.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
None of these chefs exist, as they're all AI-generated images. This hasn't stopped them from racking up lots of engagement on Facebook by posting AI-generated images of food (and occasional thoughts and prayers), however.
cc: @ZellaQuixote
These "chefs" are part of a network of 18 Facebook pages with names like "Cook Fastly" and "Emily Recipes" that continually post AI-generated images of food. While many of these pages claim to be US-based, they are have admins in Morocco per Facebook's Page Transparency feature.
Between them, these 18 Facebook "chef" pages have posted AI-generated images of food at least 36,000 times in the last five months. Not all of the images are unique; many have been posted repeatedly, sometimes by more than one of the alleged chefs.
Can simple text generation bots keep sophisticated LLM chatbots like ChatGPT engaged indefinitely? The answer is yes, which has some potentially interesting implications for distinguishing between conversational chatbots and humans.
For this experiment, four simple chatbots were created:
• a bot that asks the same question over and over
• a bot that replies with random fragments of a work of fiction
• a bot that asks randomly generated questions
• a bot that repeatedly asks "what do you mean by <X>?"
The output of these chatbots was used as input to an LLM chatbot based on the 8B version of the Llama 3.1 model. Three of the four bots were successful at engaging the LLM chatbot in a 1000-message exchange; the only one that failed was the repetitive question bot.
The spammers behind the "Barndominium Gallery" Facebook page have branched out into AI-generated video and started a YouTube channel with the catchy name "AY CUSTOM HOME". The results are just about as craptastic as you'd expect.
In this synthetically generated aerial video of a (nonexistent) barndominium under construction, the geometry of the roof changes, a blue building appears, and a tree vanishes, all in the course of just three seconds.
This AI-generated barndominium features a long AI-generated porch with some chairs on it. Exactly how many chairs there are depends on what angle you look at it from, however, as the chair on the left splits into three chairs as the camera pans.
Some observations regarding @Botted_Likes (permanent ID 1459592225952649221)...
First, "viral posts which don't result in follower growth and have very little engagement in the reply section" is not a useful heuristic for detecting botted likes. Why not?
cc: @ZellaQuixote
"Viral posts that do not result in follower growth" is not a valid test for botting, because posts from large accounts often go viral among the large account's existing followers but do not reach other audiences, resulting in high like/repost counts but little/no follower growth.
"Very little engagement in the reply section" doesn't work for multiple reasons (some topics spur debate and some don't, some people restrict replies, etc)
Hilariously, @Botted_Likes seems to be ignoring their own criteria, as many of the posts they feature have tons of replies.
As with the banned @emywinst account, the @kamala_wins47 account farms engagement by reposting other people's videos, accompanied by bogus claims that the videos have been deleted from Twitter. These video posts frequently garner massive view counts.
@Emywinst @kamala_wins47 The operator of the @kamala_wins47 account generally follows up these viral video posts with one or more replies advertising T-shirts sold on bestusatee(dot)com. This strategy is identical to that used by the banned @emywinst account.
What's up with all these similarly-worded enthusiastic posts about a Pierre Poilievre rally in Kirkland Lake, and are they all from accounts that are less than a month old? (Spoiler: yes, they are.) #Spamtastic
cc: @ZellaQuixote
An X search for "Pierre Poilievre", "Kirkland Lake", and "refreshing" performed on August 4th, 2024 turned up 151 posts from 151 accounts. All are new accounts, with the oldest having been created less than a month ago, on July 7th, 2024. (Some have since been suspended by X.)
The most intense period of activity for this group of accounts was on August 3rd, 2024, when the repetitive posts about the Poilievre rally were posted. Each account also has at least one earlier post on a random topic; some of these older posts seem to cut off abruptly.