Method note
Ledger runs on the rune bench: the method, step by step
This note expands the methodology page into a full bench session. It is the protocol behind the three-position shortlist, written so another editor could repeat it on the same builds.
Step 1 — Name the unit
A ledger run is one continuous session from the first gameplay input to the results screen. Menu time does not count. A run that ends before five seconds is logged as a warm-up, not as ranking evidence.
Step 2 — Fix the environment
Same browser family for the whole shortlist pass. Sound on. Full window, not a tiny tile. Phone passes are logged separately and used only for the input-honesty reading, not for score averages.
Step 3 — Play without ranking in mind
The first three ledger runs on a build are exploration. The desk does not decide order during exploration. Notes capture which failures were readable and which events appeared.
Step 4 — Score the four readings
Readable failure, depth of the second minute, input honesty and way back each get a short written mark, not a numeric sheet. Pipe Boost led on readable failure. Rebar Lace led on deliberate depth. Egg Crown led on on-ramp calm. Way back passed for all three because Close is present in every state.
Step 5 — Set the order once
Order is set in a single pass after the marks exist. It is not revised after the reviews are written. That rule stops the desk from rewriting the shortlist to match nicer prose.
Step 6 — Write the reviews
Each review is step-by-step and long enough to teach the loop. Facts lines stay limited to what is true without a store: local browser build, no public rating, no install counter, checked date.
Step 7 — Publish and date
The homepage, the reviews index and the field notes all carry the same checked date. When a build changes enough to move the order, the date moves with it and the methodology page records what changed.