Machine Learning Analysis from orphAnalytics Identifies 2 Authors...
Govciooutlook

Machine Learning Analysis from orphAnalytics Identifies 2 Authors Behind QAnon Messages

By Gov CIO Outlook | Tuesday, January 05, 2021

OrphAnalytics researchers have skimmed over the whole corpus of Q posts known as ‘Q drops.’ The 4952 messages were cleaned from all content stripped of individual syntax: tables, greetings, quotations from celebrities, and messages shorter than 50 characters.

FREMONT, CA: QAnon has introduced conspiracy theories to an enormous audience. Its thousands of anonymous communications have popularized narratives such as the presence of a deep child-trafficking state. It has recently motivated a series of violent attacks and has been described by the FBI as a terrorist threat.

The Swiss business OrphAnalytics has just completed an overview of all the messages sent by Q. Its proprietary technology is designed to recognize writers of written texts. The corpus of the Q texts contained two independent signals. Its new research explores the sources and individuals behind one of the most influential conspiracy theories of modern times. "Our results very strongly suggest the existence of two different authors behind Q," says Claude Alain Roten, CEO, and Co-founder, OrphAnalytics. "Moreover, these distinct signatures clearly correspond to separate periods in time and different online forums."

Former geneticist trained at Harvard and the University of Lausanne, Roten developed his text analysis approach from computational genomics. Although traditional stylometry depends on understanding words, information, or syntax, OrphAnalytics technology is solely dependent on algorithmic analysis. It compares the frequency of character patterns to individual signatures' production, regardless of the text's context. Business researchers have provided compelling evidence on various legal issues in Europe and are working with the School of Criminal Justice at the University of Lausanne.

OrphAnalytics researchers have skimmed over the whole corpus of Q posts known as ‘Q drops.’ The 4952 messages were cleaned from all content stripped of individual syntax: tables, greetings, quotations from celebrities, and messages shorter than 50 characters. They then fed the resulting elements to their software.

The study reveals that the first cycle of Q messages distinctly carries a different signature from the others. These seminal messages will appear on the 4chan web forum from October 28 to December 2017. After that, another author took over QAnon on another site, called 8chan. The signal discrepancy is large enough to leave very little doubt about this author's swap.

"The second and longest period from Dec.1 2017 to November 13, 2020 shows a single signature with a slight evolution over time. While it is not impossible that a few other persons have mixed their voices in these +4700 messages, the signal is overall very consistent and points to a single author," says Roten.

"The next step is to contribute putting a name on QAnon by comparing these signatures to those of the usual suspects," adds Roten. "To do that, we gather and cure written material from these persons to compare it with Q messages. Recent investigations point to a handful of potential authors behind Q messages, most notably the owner of 8chan forum Jim Watkins. Tracing back the history of QAnon is important. It could help to understand how and why a baseless and outlandish theory, initially destined to a few isolated hackers, ended up having such a broad social and political impact," he notes.

Weekly Brief

Read Also