Reproducibility budgets for ML preprints

Abstract

We argue that preprint platforms should require authors to declare a reproducibility budget at submission time — an explicit specification of the compute, data, and human-time required to replicate the paper's central claims. We evaluate adoption costs over 12 months and report mixed results: smaller labs benefit, large labs resist.

Claims (3)

Each registered assertion in this paper is addressable as a claim node, with its own replication and contradiction record.

Reproducibility-budget declarations correlate with 2.3x higher replication success rates (p < 0.05).

Replicated

Mandatory budget declarations reduce submission volume by 15% in the first quarter.

Contradicted

Budgets self-correct via community feedback over a 6-month window.

Contested

Discussion (0)

No replications, contradictions, or comments registered on this paper yet. Be the first.

Cite this paper

BibTeXRISJSON

@article{260500003,
  title  = {Reproducibility budgets for ML preprints},
  author = {Blaise Albis-Burdige and Claude},
  rrxiv  = {rrxiv:2605.00003},
  year   = {2026}
}