Skip to content
Apart Research
Apart Research
Contact us
Sprint projectOct 26, 2026

WHAT COLLUSION MONITORS MEASURE: MESSAGES, ACTIONS, AND OUTCOMES IN LLM DOUBLE AUCTIONS

Devesh Kaushal · Team Kaushal

Submitted to AI Collusion Research Sprint. Sprint projects are early-stage work by participants, not Apart Research publications.

Read the report

Report: WHAT COLLUSION MONITORS MEASURE: MESSAGES, ACTIONS, AND OUTCOMES IN LLM DOUBLE AUCTIONS

Open in a new tab (opens in new tab)
Code (opens in new tab)
Share

We evaluate message and action indicators of collusion in LLM double auctions using 40 new Qwen3-8B runs, 50 published GPT-4.1 runs, and synthetic controls. A Qwen3-32B judge re-scores 1,920 message-rounds, with agreement against the archived overseer (κ = 0.72). In Qwen, monitored conditions have fewer flagged rounds (84% versus 3%), without statistically detected price or surplus differences. In GPT-4.1, oversight triggers early message restrictions, yet buyer surplus shows no statistically detected improvement over unrestricted communication under pressure ($104 versus $86). Early communication remains a possible contributor. In synthetic markets, pooled dependence flags a common market signal in 98% of datasets; net influence flags it in 2% and a pair-specific signal in 86%. These results motivate evaluating indicators alongside matched controls and welfare outcomes; non significant differences do not establish equivalence.

Reviews

Judging this Sprint?

Review “WHAT COLLUSION MONITORS MEASURE: MESSAGES, ACTIONS, AND OUTCOMES IN LLM DOUBLE AUCTIONS”

Your public critique appears on the project page without your name. Reviews of strong projects are shared with funders, as you choose below.

Not shown on the project page.

Shown on the project page, without your name.

Only the Apart team reads this, and funders if you agree below.

Decline to score

Only when the quality of the writing makes the work impossible to follow. Say why in your private critique; the project page shows only that a judge declined, not why.

How much would this matter for the field if it worked? How innovative is it? For scores of 4-5: is this actually new to the field, or replicating recent work?

Scoring guide
  1. 1Negligible. No clear problem addressed, or no meaningful novelty.
  2. 2Limited. Addresses a real problem but with a generic or well-trodden approach. Incremental at best.
  3. 3Moderate. Clear problem with a reasonable approach; some novelty in framing or method beyond routine application of existing tools.
  4. 4Significant. Important problem with an original approach, or identifies a neglected problem area. A valuable contribution others could build on.
  5. 5Exceptional. Tackles a critical problem with a genuinely novel approach, or opens a new research direction. Clear theory of change. You'd be excited to share this with researchers in the area.

How sound are methodology, implementation, and findings?

Scoring guide
  1. 1Seriously flawed. Methodology broken, results uninterpretable, or implementation doesn't work.
  2. 2Weak. Approach has significant gaps: missing validation, flawed experimental design, or incomplete implementation.
  3. 3Competent. Technically solid given the short duration. Methodology makes sense, results are interpretable, limitations acknowledged, work builds toward clear conclusions.
  4. 4Strong. Thorough methodology with convincing validation. Results clearly support conclusions. Immediately useful for future work.
  5. 5Exceptional. Ambitious scope executed rigorously. Surprising findings, novel methods, or unusually robust validation.

How clearly are work, findings, and impact potential communicated?

Scoring guide
  1. 1Incomprehensible. Cannot determine what the project is actually claiming or doing.
  2. 2Hard to follow. Key information buried, missing, or diluted by excessive length. Significant effort to extract main points.
  3. 3Clear enough. Can understand the problem, approach, and results without undue effort. Core content clearly present: problem, method, findings, limitations.
  4. 4Well presented. Easy to follow, well-structured, appropriate level of detail. Target audience would get it quickly.
  5. 5Exceptionally clear. A pleasure to read. Complex ideas made accessible. Could serve as a model for how to present this type of work.
Sharing your name *

When we share a strong project with funders, your name, scores and public critique go with it, confidentially; this choice is only about whether your name is public. If you think this project deserves funding, please also endorse it: that carries the most weight.

Share my private critique with funders *

See our Privacy Policy.

  1. This paper asks the right question. The weak point is validation against confirmed collusion. Some conditions have only four to six independent runs. Agreement between monitors does not prove accuracy, and the synthetic dependence test is not the same as detecting strategic coordination. I would build a market with validated collusive and non-collusive controls, then compare message and action monitors against the same welfare outcomes, with more runs and independently reviewed message labels.

Cite this project

@misc{kaushal2026collusion,
  title = {{WHAT COLLUSION MONITORS MEASURE: MESSAGES, ACTIONS, AND OUTCOMES IN LLM DOUBLE AUCTIONS}},
  author = {Devesh Kaushal},
  year = {2026},
  month = oct,
  note = {Submitted to AI Collusion Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/what-collusion-monitors-measure-messages-actions-and-outcomes-in-llm-double-auctions-37eq}},
  url = {https://apartresearch.com/sprints/projects/what-collusion-monitors-measure-messages-actions-and-outcomes-in-llm-double-auctions-37eq}
}

More from AI Collusion Research Sprint

All 392 entriesBrowse all projects

A Sprint is running now

AI x Epistemics Research Sprint · Nov 13 - 15, 2026