Blockchain Effort Benchmark: A Verified, Commit-Traced Dataset of Actual Blockchain Software Development Effort

A reproducible, commit-traced benchmark dataset of actual blockchain software development effort, built to calibrate and test COCOMO-style software cost estimation models for blockchain projects. 44 independently verified calibration points (73 raw source rows before dependent-window collapse of repeated measurements of the same project). Candidates were screened from four contributing sources (the Web3 Foundation Grants Program, the aeternity community forum, Polkadot OpenGov and treasury governance records, and Zcash Community Grants) plus two further sources that were fully screened and correctly produced zero eligible rows (Crust Grants and the Polkadot Open Source Developer Grants program, both because their delivered projects report only planned effort, not the actual effort this dataset requires). Each point pairs a reported actual development-effort figure with an independently measured code size (KSLOC, reuse-adjusted) of the exact delivered repository at the exact delivered commit. Constructed via a PRISMA-style identification, screening, eligibility, and inclusion pipeline, with the same admission rule applied to every candidate regardless of source, and raw sources frozen with SHA-256 checksums for independent verification.