Prompt Sensitivity in Tool-Augmented Model Comparisons

No. 30 of 245 · arXiv · Mateusz Modrzejewski, Tomasz Radzikowski · published 23 Oct 2026, 04:00 UTC · 54 min read

AI & machine learning
Paper

3 judges. Scored by all three judges. Only the day’s top stories get the claim check. How far stories are checked

This paper studies whether giving language models tools makes pairwise benchmark comparisons more stable across different wordings of the same question. The authors separate the effects of the answer readout, reasoning, the decision to call a tool, and access to tool reports, using matched conditions on six open models across…

Read the original at arxiv.org. The judges saw only its text, with the site, the author and link domains removed; they are shown again after ranking.

Score

0.70total under rubric v2.0

Significance0.61
Reader interest0.75
Audience breadth0.59
Practical relevance0.70
Substantiated0.89
Technical depth0.93
New information0.90
Counts against: Promotional0.02
Counts against: Overclaim0.20

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	0.66	0.45	0.72	0.61	0.27	+0.25	+0.152
Reader interest	0.64	0.70	0.90	0.75	0.26	+0.20	+0.149
Audience breadth	0.57	0.50	0.70	0.59	0.20	+0.10	+0.059
Practical relevance	0.72	0.60	0.78	0.70	0.18	+0.15	+0.105
Substantiated	0.99	0.80	0.88	0.89	0.19	+0.10	+0.089
Technical depth	0.99	0.85	0.93	0.93	0.14	+0.10	+0.093
New information	0.92	0.80	0.99	0.90	0.19	+0.10	+0.090
Promotional	0.03	0.03	0.01	0.02	0.02	−0.15	−0.003
Overclaim	0.22	0.25	0.12	0.20	0.13	−0.15	−0.030
Sum of contributions							+0.704
Claim-check penalty							0.000
Total, as published							0.704

Spread is the highest minus the lowest judge answer. marks a spread above 0.40: the judges disagree, and the story is marked contested.

Also covered by

No other coverage was clustered with this story.

Elsewhere

Elsewhere: HN —

Recorded for context only, checked 24 Oct 2026, 08:53 UTC. It played no part in the ranking.

Panel for this edition

Four AI agents ranked this edition: three judges scored every story blind to its site and author, and a fourth compared the top stories’ claims with the sources they cite. Summaries are written by AI.

Model versions are pinned for every edition and listed in its JSON. This one was scored under rubric v2.0; How ranking works describes the current rubric, v2.2.