Puppet Dialectic

From Cognitive Attack Taxonomy

Puppet Dialectic

Short Description: Two or more sock-puppet personas, secretly run by one actor, stage an apparent debate from opposing sides in order to move the audience's sense of the reasonable middle toward a position the actor chooses.

CAT ID: CAT-2026-019

Layer: 8 (CAT v1.0); Layer III (Network) (CAT v2.0)

Operational Scale: Multiple Scales

Level of Maturity: Observed in the Wild

Category: TTP

Subcategory: Sock-Puppetry

Also Known As: Locke–Demosthenes Gambit; staged opposition; paired sock puppets; controlled opposition

Description:

Brief Description:

Closely Related Concepts: Overton window; compromise (decoy) effect; controlled opposition

Mechanism: Audiences tend to locate the "reasonable" position halfway between the visible poles of a debate, and they read disagreement between two voices as proof that both are independent. By controlling both poles, the attacker controls where the apparent middle sits. The more extreme persona works as a decoy that makes the target position look moderate, and the staged disagreement hides the coordination.

Multipliers: Personas with established followings built before the operation; a topic on which the audience has no settled view; platforms that reward conflict with reach; anonymity norms that make coordination hard to detect; generative AI that lowers the cost of running many convincing personas.

Detailed Description: The attacker builds two or more personas that appear to belong to opposing camps. Each persona argues its side sincerely and attacks the other in public. The personas' real purpose is shared: either to pull the conversation toward a position chosen in advance, usually the "sensible compromise" voiced by the more moderate persona, or to drive the poles apart so that the audience polarizes. Because the debate looks adversarial, observers do not suspect a single author, and each persona can recruit genuine followers who then carry the framing forward without knowing its source. This differs from ordinary Sock-Puppetry and Astro-Turfing, in which the fake voices all push the same side: here the deception lies in the apparent conflict itself.

INTERACTIONS [VETs]:
Sock-Puppetry (TTP, CAT-2022-140)
Astro-Turfing (TTP, CAT-2022-122)
Spectrum of Allies (TTP, CAT-2022-228)
Reflexive Control (TTP, CAT-2022-314)
Coalition Wedge (TTP, CAT-2026-003)
Decoy Effect (Exploit, CAT-2022-051)
Anchoring (Vulnerability, CAT-2022-037)
Framing Effect (Vulnerability, CAT-2022-063)
Social Proof (Vulnerability, CAT-2022-226)
False Consensus Effect (Vulnerability, CAT-2022-058)
Ingroup Bias (Vulnerability, CAT-2022-203)

Examples:

Use Case Example(s): In Orson Scott Card's novel Ender's Game (1985), the siblings Peter and Valentine Wiggin write online as "Locke" and "Demosthenes". Demosthenes takes a hard-line, populist stance while Locke plays the measured statesman; the two appear to be rivals, but Peter coordinates both voices so that Locke's position becomes the reasonable middle ground and gains political influence. A real-world operator could run the same pattern with two accounts on a policy forum: one pushing an extreme proposal, the other "reluctantly" endorsing a softer version that was the goal all along.

Example(s) From The Wild:

  • Houston, Texas, 21 May 2016: two Facebook groups run by Russia's Internet Research Agency, "Heart of Texas" and "United Muslims of America", organized a protest and a counter-protest at the same time and place outside an Islamic center. Real people turned up on both sides. The operation was disclosed by Senator Richard Burr at a Senate Intelligence Committee hearing in November 2017 (Vice, 2017; Texas Monthly, 2018).
  • Internet Research Agency accounts embedded themselves in both the left-leaning and right-leaning clusters of the 2016 #BlackLivesMatter conversation on Twitter, tailoring content to each side to sharpen the divide (Arif, Stewart & Starbird, 2018).

Comments:

The documented real-world cases show the polarizing variant, in which both poles are pushed apart. The steering variant shown in Ender's Game, in which the paired personas converge on a pre-chosen middle position, is the more precise form of the technique; a verified real-world instance of that variant would strengthen this entry. Entry added in the CAT v2.0 revision (November 2026).

References:

Vice, "Russian Facebook trolls got people to protest against each other in Texas" (2017): https://vice.com/en/article/3kvvz3/russian-facebook-trolls-got-people-to-protest-against-each-other-in-texas
Texas Monthly, "Russians sowed divisions in Texas politics, says U.S. Senate report" (2018): https://www.texasmonthly.com/news-politics/russians-sowed-divisions-texas-politics-says-u-s-senate-report/
Tech Policy Press, "Researchers reverse-engineer 2016 Texas protest organized by Russian Internet Research Agency": https://techpolicy.press/researchers-reverse-engineer-2016-texas-protest-organized-russian-internet-research-agency
Arif, A., Stewart, L. G., & Starbird, K. (2018). Acting the Part: Examining Information Operations Within #BlackLivesMatter Discourse. Proceedings of the ACM on Human-Computer Interaction 2 (CSCW), Article 20: https://par.nsf.gov/servlets/purl/10096590
Card, O. S. (1985). Ender's Game. Tor Books.
Mackinac Center for Public Policy, "The Overton Window": https://www.mackinac.org/OvertonWindow
Simonson, I. (1989). Choice Based on Reasons: The Case of Attraction and Compromise Effects. Journal of Consumer Research 16(2), 158–174.