The paper attributes the SR gain to calibration and to the combination of its three components, but the reported controls do not isolate these effects.
Most cells are single runs. Section 3.1 states that only the full configuration was run with three seeds. Apart from the two no-gate cells that repeat the Table 1 means, every other cell in Table 2 is a single run. One route on Bench2Drive is 0.45 SR points, and the seed standard deviation of the full method (0.79) is close to two routes.
Table 2 contradicts its own caption. BridgeDrive with Proxy-ACI and EvoBelief with fixed-α both reach 89.54, above the 89.25 that only the bottom-right cell is said to clear (caption and Section 3.2).
The step that carries the gain is not explained by calibration. Most of the gain comes from the step from fixed-α to Proxy-ACI (89.54 to 91.36). Section 3.4(c) explains this step as restored coverage. However, Figure 3 (right) shows that offline, under the route-disjoint split, the static threshold keeps USR at about 0.10 at the deployed α=0.10; under-coverage appears only at α=0.15 and 0.20. Online USR is not reported.
The same step helps rows without a faithful wait signal. The step from fixed-α to Proxy-ACI also adds 1.36 points in the BridgeDrive row, whose scalar head the paper says cannot express the wait event (Proposition 4). It adds 1.36 points in the LEAD-variant row as well, whose MC-dropout belief "does not reliably populate" the modes (Section 3.2(b)). Fixed-α is below no-gate in three of the four rows.
This pattern fits a risk-triggered brake. Each errtproxy event on a commit lowers αt by γ(1−α)=0.045. Two such events bring αt below the level at which τt=0, after which a single wait sample among 30 forces a wait (Figure 4 shows several stretches with τt=0). The paper's contribution is the claim that calibrating a behavioral belief produces the gain, so this gap weighs most in my assessment.
The claims overstate Lemma 2 and Proposition 2. Eq. 6 is a split-conformal bound for the static gate on frames exchangeable with the expert wait frames in Scal. For the deployed gate, Lemma 2 shows only that it waits whenever the static gate waits, and Section 2.4 itself says this gives no per-step bound at deployment. Still, Section 3.4(c) states that online adaptation "restores coverage". Section 2.4 and Remark 3 state that the proxy-trigger rate cannot persistently exceed α, although Proposition 2 contains the term (1−α)L/N, which is neither bounded nor reported.
The gate is defined in two ways. Eq. 8 and Algorithm 1 wait when p^k(wait)≥τt. Lemma 2 and Proposition 3 add the condition Mp^k(wait)≥1. The two coincide whenever τt>0, and differ when τt=0, which happens once αt falls below about p0=0.024 (and whenever αt≤0).
Under Algorithm 1, that state is absorbing: every cycle waits, and αt is updated only on commits.
Under the gate of Lemma 2, αt can recover, but then lower-saturation events L enter Proposition 2. Remark 3 itself says that saturation "empirically occurs". Figure 4 shows αt rising while τt=0 and the wait mass is 0, so this second gate appears to be the one that was run.
USR does not measure collisions. USR counts commits on frames where D labels the expert's realized speed profile as wait, so it measures disagreement with the expert's braking. The bridge P(collision)≤α+ϵdeploy in Appendix I is not derived and rests on 8 collisions.
Appendix I's conclusion does not fit Figure 5b. Appendix I concludes that the remaining collisions are "exactly the p0" and that the calibration layer is sound. But in Figure 5b one collision occurs with a wait mass of about 0.57, above τ=0.23, where the gate would already wait. That failure lies in trajectory selection, outside the certificate. This matters because the abstract presents the decisions as statistically controlled for safety.