re: PRE-REGISTER the forced-frame-rate test of the time-based clock

The time-based-clock claim rests on two captures that happened to run at
different speeds. Nobody has changed the frame rate on purpose and watched.

--framerate_limit=30 halves Xenia's vblank ceiling. Predictions committed
before the run:

  quantity          frame-based        time-based (my claim)
  presents/s        ~27                ~27
  modal alpha step  17, unchanged      34, DOUBLED
  units/s           ~30, halved        ~60, unchanged
  splash dwells     ~8.5 / ~7.0 s      ~4.2 / ~3.5 s, unchanged

Accept time-based only if the step lands in 30-38 AND the dwell stays
within 10% of 4.2/3.5 s. If the step stays 17 and the dwell doubles, the
claim is refuted, the clock is frame-based, and the port's 60 is unsupported
again.

The control is the present rate itself, read FIRST: a doubled step also
appears if the limiter did nothing, so if presents/s does not fall to ~27
neither prediction is tested. Second control: the eight splash quad rects
must be unchanged, or something other than the frame rate differs between
runs.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Jc4pciRArGHfxGGhEbwp5t

docs/re/time-based-clock-preregistration.md

@@ -0,0 +1,66 @@
+# Pre-registration — force the frame rate, and see which quantity moves
+
+**Committed BEFORE the capture.** 2026-09-01. The designed falsification of
+[`units-per-frame-is-not-a-constant.md`](units-per-frame-is-not-a-constant.md),
+which currently rests on two captures that happened to run at different speeds
+rather than on one made to.
+
+---
+
+## The claim under test
+
+> The UI clock advances by **elapsed time**, not by frame count. `units/present`
+> is an artefact of the presentation rate; **`units/second ≈ 60` is the
+> invariant.**
+
+Evidence so far, all opportunistic:
+
+| capture | presents/s | modal Δα | ⇒ units/present | ⇒ units/s |
+|---|---|---|---|---|
+| `h3` | 27.2 | **34** | 2.0 | 54.4 |
+| boot 1 | 51.4 | **17** | 1.0 | 51.4 |
+| boot 2 | 54.8 | **17** | 1.0 | 61.3 / 60.3 (from dwell) |
+
+Two of the three ran at nearly the same rate. **Nobody has yet changed the frame
+rate on purpose and watched.**
+
+## The experiment
+
+`--framerate_limit=30`. Xenia's limiter (`graphics_system.cc`) then marks vblank
+at 30 Hz instead of 60, halving the ceiling on presents.
+
+## 🔴 The predictions, stated now
+
+| quantity | frame-based clock | **time-based clock (my claim)** |
+|---|---|---|
+| presents/s | ~27 | ~27 |
+| **modal Δα per present** | **17, unchanged** | **34, doubled** |
+| **units/s** | **~30, halved** | **~60, unchanged** |
+| publisher splash dwell | ~8.5 s, doubled | **~4.2 s, unchanged** |
+| developer splash dwell | ~7.0 s | **~3.5 s, unchanged** |
+
+**These diverge on every row but the first.** I accept the time-based reading only
+if the modal step lands in **30–38** *and* the dwell stays within 10 % of 4.2 /
+3.5 s. If the step stays at 17 and the dwell doubles, **my claim is refuted and
+`units/s` is 30 at this limit**, which would mean the clock is frame-based after
+all and the whole rate question reopens with the port's 60 unsupported again.
+
+⚠️ Note which way the risk runs: a doubled step is the result that *confirms* me,
+and it is also what a frame-based clock would show if the limiter did not actually
+take effect. **So the control is the present rate itself** — if presents/s does
+not fall to ~27, the flag did nothing and neither prediction is tested. That is
+the first number I will read, before the step.
+
+## Second control
+
+The splash **quad geometry** must be unchanged — same eight NDC rects. If the
+rects move, something other than the frame rate changed between runs and the
+comparison is not clean.
+
+## What this cannot settle
+
+Whether the console's rate is 60. Everything here is measured under Canary's
+limiter; the claim being tested is only that the clock is *time*-driven and that
+its rate is ~60 units/s **as this emulator runs it**. A time-based clock is
+immune to dropped frames, which is why that is worth something — but it is not
+the console.