Near-replicates
The rare gift in this data: two or more hits that used almost the same attack while the defender’s city had changed (heal, reins gone, a layer missing, a reloc). That is the closest thing to an experiment. The usual gift is the opposite — you changed recipe and the wall changed, and nobody can say which matter.
By hand you notice a near-replicate only if you were staring at both reports’ layer screens. You do not notice it from opening shots.
Focus
One keep, attacker or defender vantage. You need the sequence first. Then ask for the pairs that look the same on our side.
Why this used to die as a daydream
“Same rally, different result” is the argument every setter has. Without lined-up compositions and the gap between hits, it is folklore.
Questions (any wording)
- Which hits on this keep were nearly the same attack into a different wall?
- What had changed on their side between those two — heal, reins, mix?
- Did we keep sending the same recipe after the thing it was for was already gone?
- On the teammate city: which incoming looked like a repeat while our reins had changed?
What a good answer looks like
A pair (or cluster) of hits, what was similar on the attack, what differed on the city, and a refusal to call the better result “optimal play” from two rows. One keep is the parent family.
Common miss
Calling any two ranged rallies a replicate. Near means composition and formation, not “same troop type word.”