Event to Loop

When a fix works briefly and the problem returns, trace one consequence back to an earlier cause and mark its sign and delay.

After every big blowup with your brother, you both apologize, things warm up for a few weeks, and then the same fight returns wearing slightly different clothes. Each round gets treated as an event with a cause, this comment, that slight, but events that reliably regenerate aren't really events. Something about the resolution is feeding the next occurrence.

Event to Loop is a 30-second trace for exactly this cue: a fix works briefly and the problem returns. The procedure: pick one consequence of the last episode and trace it backward to an earlier cause, then mark two properties: the sign (does the connection amplify the problem or dampen it?) and the delay (how long before it feeds back?). Here: after each fight you both avoid real topics (consequence), avoidance lets grievances accumulate unspoken (feeding an earlier cause), and the accumulation erupts weeks later. That's a reinforcing loop with roughly a one-month delay, and it explains why apologies, which treat only the eruption, never hold. It works when the recurrence pattern suddenly makes structural sense and the intervention shifts to breaking the avoidance link. The spec's warning: recurrence alone is not proof of feedback. Some things just keep happening for external reasons; demand an actual traced connection before declaring a loop.