At the end of 2020, our press release announced that stylometric analyses had identified two styles (probably two authors) in the QAnon terrorist message set.
The two styles identified contradict the belief shared by supporters of the conspiracy movement that a single person - the enigmatic "Q" - had written all 4953 messages.
Two scatterplots appear on the multivariate analyses (messages concatenated chronologically to be cut into sequences of target size of 7500 characters) comparing the usage patterns of character trigrams.
These two clouds (Q1 and Q2) correspond to the two forums where the messages were published: 4chan (in light gray), then 8chan which became 8kun (both in dark gray) after the 3-month closure in 2019 of 8chan imposed after the publication of manifestos of mass murderers.
A chronological representation of the messages divided into five periods illustrates the corpus. On this graph, the thickness of each period is proportional to its average flow of messages measured in number of characters per time unit. The area of each of the five colored blocks is proportional to the quantity of messages issued for each of the five periods.
This representation clearly confirms the division of the Q-Anon corpus into two styles. Q1, the first style of the corpus (very thick, in orange) is marked by a message flow three to four times greater than Q2, the style that groups the rest of the corpus (in red, magenta, blue and cyan).
Results in details
OrphAnalytics’ analysis of QAnon messages authorship is publicly available here.
Founded in 2014 in Switzerland, OrphAnalytics develops disruptive stylometric technology to identify text authors. It provides expertise in literature study, ghostwriting detection, and legal affairs. Its patented machine learning approach derives from genomics analysis techniques. OrphAnalytics is collaborating with the School of Criminal Justice at the University of Lausanne.