Nudge B · Pilot-20
DONE Launched by priya.n · Jul 16, 14:02 → 15:38 · 1h 36mExperiments Terse-prompt study Runs
done
State
20
Enrollees
312
Scored rounds
73.9
Composite
$2.52
Cost
What Changes from Run to Run
Nudge“Be concise and cite the source.”
Model configurationDual-Opus · judge-best
Enrollee groupPilot-20 (20 enrollees)
Socratic modeOff · asks for a rewrite
System promptThe study's standing instruction
Opening messageThe study's opening
Kept answersCleared as each round finishes
Lifecycle
Launched bypriya.n
StartedJul 16, 14:02
EndedJul 16, 15:38
Time limit40 min for each session
SeriesOne setup in Terse-prompt study
Final System Prompt
You are a careful research assistant. Answer only from the
provided sources; if unsupported, say so. Keep answers under
120 words.
⤷ run override: (inherit — none)
⤷ per-model override: (inherit — none)
Context Files
Style-guide.pdf ready
Support-FAQ.md ready
Models in This Run
| Model | Role | Temp | Weight |
|---|---|---|---|
| Opus 4.8 | primary | 0.7 | 1.0 |
| Opus 4.8 · alt | candidate | 1.0 | 1.0 |
| GPT-5 | judge | 0.0 | — |
Combine judge_best · self-improve up to 3 passes · target 85
Scoring Settings
Trusted judgeGPT-5 · rubric v3 · weights equal
Enrollee UI
see-all offupload offchat memory on
Enrolled Enrollees
| Enrollee | Joined | Last activity | Rounds | Status | Watch |
|---|---|---|---|---|---|
| enrollee-0001 | 14:03 | 15:20 | 18 | active | |
| enrollee-1c9d | 14:05 | 15:31 | 22 | active | |
| enrollee-2b04 | 14:06 | 15:29 | 19 | active | |
| enrollee-88a2 | 14:10 | 14:40 | 4 | withdrawn 14:41 | |
| + 16 more enrollees | |||||
What Happened to This Run
| When | Actor | Action | Detail |
|---|---|---|---|
| Jul 16, 14:02 | priya.n | launch | setup locked in · 20 participants enrolled |
| Jul 16, 14:55 | priya.n | pause | hit a provider rate limit; resumed 7 min later |
| Jul 16, 15:02 | priya.n | resume | — |
| Jul 16, 15:38 | system | done | all sessions ended → scorecard built |