NUWA Risk Evaluation · Method
Impact of modeling assumptions on time horizon results
NUWA research note · independent review pending
NUWA Research Note
Research questionDesign and assumptionsSource findingsNUWA interpretationLimits and contrary evidenceResearch and decision relevanceEvidence updateInstitutions and rolesOriginal resourcesModel-assisted reading · Selected sections of the original full text or official article; a complete section-by-section audit remains pending.
At a glance
Saturation amplifies sensitivity to model assumptions; time-horizon revisions are not capability regressions.
Research question
How sensitive are time-horizon estimates to statistical assumptions near saturation?
Design and assumptions
The authors’ research note examines regularisation corrections, fitted curves, task distribution and human-duration noise, distinguishing sampling uncertainty from model-choice sensitivity.
Source findings
Some frontier horizon estimates change materially under plausible modelling choices, particularly as the task suite approaches saturation. A methods correction is not an observed model capability regression.
NUWA interpretation
Each horizon estimate needs its method version and task distribution. Error bars do not automatically cover model-choice uncertainty, and extrapolation is not observed continuous autonomous runtime.
Limits and contrary evidence
The research note has less review than a formal METR blog. Conclusions concern the evaluated task distribution and fitting assumptions.
Research and decision relevance
NUWA recommendation: Use for version and sensitivity notes on the existing Time Horizons record; do not overwrite historical observations.
Precise source locations
- Impact of modeling assumptions on time horizon results · Summary; model correction; uncertainty table; text lines 00043,00069,00240Original resource captured 2026-10-28T07:24:45.348612+00:00; sha256:9695043f6347be4a67091843e5b34b1db63f1ab11ee6a7956edcf1094182bd20
Source checking, extraction, editorial approval and replication are recorded separately. Methodology →
Evidence update · 2026-10-28
Model-assisted reading · independent evidence and bilingual editorial review pending. Values are source findings or inventory counts, separate from the trend result series.
Related reading
NUWA reading connections; these do not imply citation, replication or score equivalence.
Institutions and roles
Cite and reuse
Cite the original publication for its findings. Cite this NUWA record separately for its explanation and retained snapshot.
Original publication · BibTeX
@misc{horizon-assumptions,
title = {Impact of modeling assumptions on time horizon results},
url = {https://metr.org/notes/2026-03-20-impact-of-modelling-assumptions-on-time-horizon-results/},
year = {2026},
date = {2026-03-20},
version = {Original resource captured 2026-10-28T07:24:45.348612+00:00; sha256:9695043f6347be4a67091843e5b34b1db63f1ab11ee6a7956edcf1094182bd20},
note = {Original resource captured 2026-10-28T07:24:45.348612+00:00; sha256:9695043f6347be4a67091843e5b34b1db63f1ab11ee6a7956edcf1094182bd20}
}
NUWA record · BibTeX
@misc{horizon-assumptions-nuwa,
title = {Impact of modeling assumptions on time horizon results — NUWA research record},
url = {https://whitzard.tech/en/nuwa/whitzard-index/work/horizon-assumptions/},
version = {2026-10-29.1},
organization = {NUWA Frontier AI Safety Lab},
howpublished = {Retained note edition: https://whitzard.tech/nuwa/whitzard-index/data/record-editions/2026-10-29.1--horizon-assumptions.json},
note = {NUWA research note. Source-reading scope and independent review status are stated on the record. Snapshot 2026-10-29.1. Retained note edition: https://whitzard.tech/nuwa/whitzard-index/data/record-editions/2026-10-29.1--horizon-assumptions.json.}
}
Evidence snapshot 2026-10-29.1 · Index updates · RSS
References
- [1] Impact of modeling assumptions on time horizon results. 2026-03-20 ; Summary; model correction; uncertainty table; text lines 00043,00069,00240