Thinking Machines Lab's research-as-product model ($2B, 89% MoM growth) will be replicated by 2-3 additional companies within 18 months.

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Research-as-product company model works at venture scale now, proven by Thinking Machines Lab.claim0.805
• If ReflectiveBench receives 5+ citations by month 12, About Blank will prioritize Lab-with-frontier-partnership over independent Lab structure.prediction0.733
• Thinking Machines Lab's interaction-model discourse positioning leaves UI/invocation-mode layer open for Company-shape competitor.prediction0.725
• One-stage CoT (QCM→RA) shows 12.31% accuracy drop vs. no-CoT (QCM→A) on ScienceQA; two-stage framework (rationale generation + answer inference) achieves 85.31% accuracy with vision featuresfinding0.723
  Empirical evidence that naive one-stage CoT fails in language-only setting; two-stage + vision achieves state-of-the-art.
• About Blank will add P5 (Company-shape probe) and P6 (ReflectiveBench-as-glue spec) to PRONGS.md within 7 days.prediction0.722
• Cognitive glue framework (Levin-Lyons) will be adopted as standard language for analyzing collective intelligence architectures in AI, economics, and organizational design within 18 months.prediction0.720
• The 30-coin thought experiment shows the step-by-step approach works (about one minute) while the all-or-nothing approach does not (about 300 years), and this is the secret of biological evolution.claim0.720
  Mathematical argument that adaptation without stepwise process is statistically impossible.
• Nearly all of 64 simulated agents attain 100% performance at around the 10th trial when allowed to perform abductive Bayesian model reduction after each trial.finding0.719
  Group-level simulation result showing generalizability of BMR benefit across agents