MIT paper: stating rules beats reasoning prompts for LLM agents — Gemma's underbid gap shrinks from $5.30 to $0.30
rohanpaul_ai · x · 2026-10-06
- A new MIT paper studies LLM agents in auctions and matching games: choices presented as simple steps or plainly stated safe moves improve decisions, while prompts to reason about opponents hurt.
- Market-design rules of thumb built for humans carry over to LLM agents; honest bids and rankings are always optimal.
- GPT-4o, Claude, Gemini and Gemma still underbid to keep profit margins.
- A rising price with a stay-or-exit choice moved Gemma from $5.30 below its value to $0.30 below; a one-line note that rejections only redirect cut matching errors from 4.2% to 0.2%.
- Practical takeaway: fix format and state key facts before adding reasoning prompts; judge agents by choices, not written plans.
More from coding & agent
- Granola now auto-captures screen shares in meetings, not just what's said — soleio · 2026-10-06
- Spawn turns Discord feature requests into shipped features in under 12 hours — majidmanzarpour · 2026-10-06
- Used OpenAI Dots as a Free Agent Swarm to Break a 47-Year-Old Math Record — jaxchang · 2026-10-06
- Build an Internal Company Chatbot by Wiring Grok Bot to Slack and Your SOPs — alexcovo_eth · 2026-10-06
- Dev shares semantic router results: guard model catches 87% of attacks — colinmcnamara · 2026-10-06
- Graphical launches: a design tool for visual languages that pairs with coding agents — mikelikesdesign · 2026-10-06