Precision-based sizing finds the number of pairs that gives a confidence interval on the McNemar paired-proportion difference a target half-width.
McNemar's effect size is driven by the discordant-pair count, so the SE depends on p_disc (probability of discordance) more than on the total n.
Discordance rate p_disc + signed difference p_diff + target half-width.
Required number of pairs.
Low expected discordance ⇒ required pairs balloon. Plan around the realistic discordance rate, not the desired p_diff alone.