Benchmarks

What we measured

Every figure on these pages is computed from result files committed alongside the studies that produced them, by a generator that fails the build if a claim stops holding in the data. That includes the measurements that do not flatter us.

Public reference systems Generated, not hand-typed Limits stated on every page

Available now

Study	The question	Headline
Re-analysis at zero incremental cost	When conditions change, what does re-deriving the answer cost?	Nothing, after the up-front work — on top of an exact reproduction of a published benchmark
Why not just run Monte Carlo?	Sampling is unbiased, mature and free. What can it not do?	It reports zero probability for events it never drew — and cannot tell you it failed

In preparation

Two further studies are being written up: an annual resource-adequacy analysis cross-validated against an independent open-source toolchain, and a region-decomposition program measuring how far local re-analysis scales before it stops paying.

A standing invitation. The reference systems are public and the result files are committed. We would rather you reproduced these numbers than took them on trust, and a reproduction package is in preparation.