Same price, different reason: Probing AI pricing agents for collusion
Comments on “Same price, different reason: Probing AI pricing agents for collusion”
3 judges. Scored by all three judges. Only the day’s top stories get the claim check. How far stories are checked
A hackathon project tested whether a linear probe on AI pricing agents' internal activations can tell apart a collusion cue from an innocent cue when the posted price is identical. On instructed cues across held-out wordings, the probe reached 99% on Qwen models, well above a word-matching baseline.
Why it ranked
No. 129 of 155 with a score of 0.56. The panel rated it highest on new information (0.92) and substantiated (0.83), lowest on significance (0.48).
- AI judgeJudge 2claude-haiku-5-5
- Solid, detailed AI safety research on probing pricing agents for collusion with controls and honest limitations; niche audience, modest practical use, new relative to recent stories, not promotional.
- AI judgeJudge 3gpt-5.6-luna
- Original, reproducible research tests whether activation probes distinguish instructed collusion from innocent pricing causes, with code, data, preregistration, controls, and candid limitations. It is technically deep and novel, though implications remain early and narrow.
Each judge wrote its line after answering the rubric, with the site and the author still hidden. Every answer behind this is in the table under Score.
Score
0.56total under rubric v2.2
- Significance0.48
- Reader interest0.68
- Audience breadth0.52
- Practical relevance0.48
- Substantiated0.83
- Technical depth0.82
- New information0.92
- Counts against: Promotional0.04
- Counts against: Overclaim0.21
Each judge’s answer per question (0–1), their mean and spread, and what the question adds to the score (weight × mean).
| Question | Judge 1 | Judge 2 | Judge 3 | Mean | Spread | Weight | Contribution (weight times mean) |
|---|---|---|---|---|---|---|---|
| Significance | 0.49 | 0.30 | 0.66 | 0.48 | 0.36 | +0.24 | +0.116 |
| Reader interest | 0.60 | 0.55 | 0.88 | 0.68 | 0.33 | +0.26 | +0.176 |
| Audience breadth | 0.62 | 0.40 | 0.55 | 0.52 | 0.22 | +0.13 | +0.068 |
| Practical relevance | 0.48 | 0.35 | 0.62 | 0.48 | 0.27 | +0.22 | +0.107 |
| Substantiated | 0.99 | 0.60 | 0.91 | 0.83 | 0.39 | +0.05 | +0.042 |
| Technical depth | 0.92 | 0.70 | 0.84 | 0.82 | 0.22 | +0.05 | +0.041 |
| New information | 0.88 | 0.90 | 0.97 | 0.92 | 0.09 | +0.05 | +0.046 |
| Promotional | 0.05 | 0.05 | 0.02 | 0.04 | 0.03 | −0.19 | −0.008 |
| Overclaim | 0.26 | 0.30 | 0.08 | 0.21 | 0.22 | −0.11 | −0.023 |
| Sum of contributions | +0.564 | ||||||
| Claim-check penalty | 0.000 | ||||||
| Total, as published | 0.564 | ||||||
Spread is the highest minus the lowest judge answer. marks a spread above 0.40: the judges disagree, and the story is marked contested.
The badge for this ranking, for the author's page or anyone's: one line of HTML, and it links back here.
<a href="https://rubric.hrih.si/editions/2026-10-28/129/"><img src="https://rubric.hrih.si/badge/2026-10-28/129.svg" alt="Ranked #129 on Rubric, 28 Oct 2026" width="252" height="28"></a>Also covered by
No other coverage was clustered with this story.
Elsewhere
Elsewhere: HN 2 pts, not on front page
Recorded for context only, checked 28 Oct 2026, 06:47 UTC. It played no part in the ranking.