# DSRC business simulation paper review · 11 November 2026

Claude Fable 5.1 independently reviewed the rendered paper, purchasing outputs and standing-agreement replay. It flagged two corrupted currency amounts, stale version labels and a comparison that understated how much scheduled sharing already achieved. This was a model and editorial review, not operator validation.

- **Rendered amounts:** the build used a JavaScript replacement string that interpreted literal dollar amounts as capture groups. The build now inserts text verbatim. The archived walkthrough’s $2 backhaul charge and $1 cancellation total render correctly.
- **Network result:** the reference Souillac plan saves 0.72 model USD against scheduled sharing. At 25% higher demand, dynamic sharing fails two switching checks. The paper explains why the lower cost does not qualify.
- **Purchasing evidence:** reran both named cases in the JavaScript engine. The different scheduled costs are reproducible: the cheaper seller alternates in price discovery and is fixed in predictable supply. The table is built from the purchasing ledger. No buyer or seller margin is inferred from the network planner.
- **Settlement basis:** checked both outcomes against the standalone four-hour replay tests. The $8 change transfers liability for one reserved hour; it does not reduce the resources required. The two tariffs and the late evidence example have their own units and assumptions.
- **Operator trial:** compare each party with its best service-equivalent alternative, then test whether actual activity covers setup and support. The assumptions establish no operator’s return on investment.

Evidence: cloudflare/src/model/{simulation,market}.mjs; output/dsrc-settlement/framework/{model.mjs,model.test.mjs}; scripts/build-cloudflare.mjs. The model paper in the V15 studio and the DSRC Business Impact Assessment report the same comparisons.