SCCR Reproduction Kit

Reproduce the Storage Cost Coverage Ratio (SCCR) yourself — in about 15 minutes, from a frozen input capture, with three independent implementations that agree to the last decimal.

Quickstart

python3 reproduce_sccr.py --input input/fee_history_capture.json
cc reproduce_sccr.c -o reproduce_sccr && ./reproduce_sccr
bash cross_check.sh   # all three must agree within 1e-6

What's in the kit

• External reproduction, step by step — the full instructions, inputs, and expected outputs.
• README — kit overview, method, and provenance.
• reproduce_sccr.py — Python implementation.
• reproduce_sccr.c — standalone C implementation (no dependencies).
• cross_check.sh — cross-checks the implementations.
• input/fee_history_capture.json — the frozen input capture.

Reference result

From the frozen snapshot (2026-09-16, 155 blocks, heights 967,138→967,292, model-spec v2.1.1) the kit reproduces avg SCCR 0.289645 (min 0.059036, max 1.294753), with 152 of 155 blocks below 1× (98.1%). The whole series rescales by 32,000/26,586 = 1.2036× from the N re-base (see the Changelog); the kit reads N from model-spec.json, so a clone reproduces the current value with no edits. JavaScript, Python, and C agree to within 1e-6.

This kit reads a frozen capture by design. The SCCR the live dashboard shows is refreshed daily by GitHub Actions and is available as a dated reading at /data/sccr.json.

Why it matters

The measurement is only as good as its reproducibility. Anyone can re-derive the number from the same inputs and check it against the published result — that is the standard this research holds itself to.

← All research · ← Back to Learn