We report an empirical, hardware-defined informational time dilation effect on superconducting quantum processors: when an operational quantum clock is locally coupled to irreversible record creation (mid-circuit measurement+reset), its inferred clock progress is reduced relative to a schedule-matched baseline, while coherent eraser (quantum-eraser style) controls remain near baseline. We define a dilation factor D N₄₅₅/N by inverting a monotone baseline curve p (S=1 A (N) ), where A (N) is a reference clock circuit and N₄₅₅ is the baseline step count reproducing the observed clock readout under a test condition. Across schedule-matched sweeps, increasing the density of irreversible records yields a strong monotone decrease in D down to D 0. 2--0. 3, whereas coherent erasure at the same interaction depth yields D 0. 93. Multiscale variants show that the temporal distribution of records matters: interleaving record creation slows the clock substantially more than blocking at matched totals. Functional-form comparisons across regimes do not uniquely select the square-root scaling proposed in the current DAGI draft; instead we propose a more robust Generalized Record-Rate (GRR) law class linking D to an empirically defined irreversibility/record-rate metric. In the latest orthogonalized local-law campaign in the C1 series on ibmfez (MAXv2; 92 circuits, 8 screened pods, 4096 shots/circuit, two randomized passes), irreversible branching remained slower than the reversible control on 87. 5% of pods in both passes, slower than a deterministic no-branch control and a random-record sham on 100% of pods in both passes, and slower than an erased-record control on 100% and 87. 5% of pods in passes 1 and 2. Baseline inversion clipping was eliminated (0. 0 in both passes) and a direct, inversion-free observable (an effective rotation-rate proxy ₄₅₅) preserved the same ordering. Model comparison across proxy families then selected an information-aware predictor () over a two-qubit load proxy by a decisive margin (=64. 183). To fulfill a promise made in the DAGI core draft, we also summarize C1-LC light-cone transport pilots: in one regime we observe strong clock dilation (D 0. 384) while causal-cone transport remains invariant within uncertainty (vₑ₄₂₎ₑ₃/vₒ₇₀₌ 1), motivating a conformal-metric refinement in which records renormalize proper time without shrinking null cones. Finally, the Tier-4 remote program now closes as a split result. In the dual-backend T4X240 package, the remote clock response remains bounded/null on both ibmfez and ibmₘarrakesh: far low-overlap shifts are tiny (D 0. 0028 and 0. 0015, respectively), both remain below 0. 1E₋₎₂₀₋, and the pooled overlap-only model outperforms a nonzero clock model (=11. 779 in favor of the simpler explanation). By contrast, a strict pulse-matched dual-backend GHZ-WR closure (GHZWRC2) shows a replicated far witness effect with strong erase restoration and a small no-entanglement add-on: on ibmfez, ₆₇ₙ=-0. 1323 (95\% CI -0. 1741, -0. 0905), R₄ₑ₀ₒ₄=0. 933, Q₍₋=0. 056; on ibmₘarrakesh, ₆₇ₙ=-0. 1569 (95\% CI -0. 1767, -0. 1371), R₄ₑ₀ₒ₄=0. 964, Q₍₋=0. 071. The remote evidence therefore does not support a universal nonlocal clock slowdown; instead it points to a dissociation between a bounded remote clock response and an active distributed coherence witness channel. Crucially, the empirical r D map (a GRR law class) extracted here provides an experimentally calibrated temporal primitive for the companion Emergent Spacetime Geometry (ESG) program, where g₀₀ is defined as a local function of the record rate r (the generalized record rate, GRR), with g₀₀ -D² in the simplest identification. In the ESG validation pipeline, these hardware primitives parameterize a 10, 000-node classical DAG simulation that generates localized time-dilation wells (inner/outer separation z 110. 7) and satisfies weak-field Einstein-surrogate closure diagnostics (median Poisson relative RMSE 0. 0029).
Petr Sramek (2026) studied this question.