OpenAlex ·

It's the Problem, Not the Path: Budget and Difficulty Confounds in LLM Reasoning Trajectories

Yiğit Utku Bulut

Published
Source
OpenAlex
Published in
arXiv (Cornell University)
DOI
10.48550/arxiv.2609.03436
Original
Read it at the publisher
Topics
Artificial Intelligence

Abstract

Reasoning traces of large language models are widely read as containing "breakthrough" moments and early-legible fates. Both readings rest on measurements missing a counterfactual control at the level of the claim; we supply both controls. First, a restart-controlled truncation probe separates when a solution fits the continuation budget from when a prefix carries value that fresh computation cannot buy, comparing per-anchor continuation solve rates against from-scratch restart curves at matched total generated-token budget. Applied to 178 problem–model cells (89 MATH problems × two small open models, an outcome-blind but difficulty-targeted cohort), exactly 1 of 178 cells survives as prefix-limited; restart dose–response separates a compute-starved model from a capability-limited one; and wherever the matched budget lies inside the restart grid, continuing the model's own prefix beats restarting (9 of 9) — predominantly compute compression rather than expanded reachability. Second, a pre-registered, difficulty-controlled test finds no detectable outcome information in early-window internal signals beyond a problem-difficulty baseline, and two generation-free analyses of public corpora show why this control is needed: a trace-blind difficulty proxy reaches AUROC 0.873 on 192K DeepSeek-R1 generations — inside the published probe range — and a closely matched reconstruction of the closest published early-window positive recovers a comparable pooled result (0.849) while within problem it is statistically indistinguishable from chance at all ten anchors (0.496 at t = 4); a post-hoc within-targeted probe finds only a small average residual, concentrated in three low-failure problems. High pooled probe AUROCs cannot by themselves establish within-attempt information; a question-only baseline or within-problem evaluation is required. Code, frozen pre-registration amendments, and all result artifacts: https://github.com/bulutyigit/problem-not-path

The authors’ abstract, as arXiv lists it. arXiv’s listings, abstracts included, are free for anyone to reuse.

Read it on ReadSearch

Each opens after you create an account, and brings you back to this paper.

Already a reader? Sign in

All new papers in Artificial Intelligence

Get the few papers worth your time in Artificial Intelligence every morning.

ReadSearch reads the new papers in your field each day and brings you the few worth reading, each with its key finding in a sentence.

Papers belong to their authors and publishers; every page here links to the original. All research topics