With perfect transmission, reversing tied rows changed 30-minute reach for 0.48% to 2.02% of active pairs in six 20-second contact records

We tested how row order changes possible paths in six public SocioPatterns contact records. The model had perfect transmission, waiting, and no recovery. In each complete 30-minute window, both people in a tested ordered pair had a contact. A sequential pass let new reach pass on within a time bin. Reversing only rows with the same native 20-second label changed 0.48% to 2.02% of pair decisions across the six records. In the high school, 44,002 of 2,865,208 decisions changed, while total reached pairs changed by only 222, or 0.00775 percentage points. Thus total reach hid most pair changes. Two independent uniform row orders disagreed on 0.27% to 1.13% of native decisions. These means use 16 independent order pairs from 32 draws. A one-hop rule and component closure gave exact bounds. Their native reach gap was 0.6 to 2.9 percentage points. With five-minute bins, it was 4.0 to 34.2 points. A boundary shift and alternative denominators changed the sizes. The package gives aggregate counts, four figures, source hashes, code, and reproduction instructions. Independent scalar checks passed. A full repeat matched count files exactly. This is a benchmark for a known implementation issue. It does not estimate actual transfer or validate a disease model.

temporal networks · human contacts · reachability · simultaneous contacts · row order · reproducibility