A calibrated readout boundary is scanned with low, tied, and high signals before count statistics. Exact arithmetic here means exact results for the stated model inputs; measured inputs still carry uncertainty and significant-figure limits.
highlighted = computed this step
A readout threshold needs two calibration levels
The zero signal is 1/4, the one signal is 3/4, and the threshold is 1/2.
szero=41,sone=43,T=21
A low signal reads zero
Signal 3/8 is below the threshold, so this finite model returns zero.
s=83<T⇒zero
Low, tied, and high signals have different decisions
The equality row is deliberately not rounded upward. At the threshold the rule says repeat, because the readout boundary is ambiguous in this finite model.
s832185decisionzerorepeatonereasons<Ts=Ts>T
A high signal reads one
Signal 5/8 is above the same threshold, so the decision changes to one. The scan makes the boundary visible before any statistics are applied.