Two review cards in a row sat unanswered (4 weeks, then 14 days) and each silently stalled its downstream. A system built for a daily operator needs defaults that execute — not just defaults that wait — before you become a weekly one.
After Aug 15 you're back to consulting rhythms: the sabbatical's daily-attention assumption breaks permanently. The Jul 26 card stall wasn't an anomaly, it was a preview — curriculum unauthored for a week, three decisions frozen, and no part of the system escalated or proceeded.
Your decision cards already carry an 'If you don't answer:' line per decision — a default. But today the default is informational: nothing reads it, nothing acts on it. The system is strictly human-in-the-loop: it stops and waits, indefinitely, for markup that may be weeks away. The stall is silent because waiting doesn't look like failure to any watcher.
The upgrade is human-ON-the-loop: after a stated expiry (say 7 days), the card's defaults execute automatically, the actions are logged, and the card moves to archive marked 'resolved by default'. You keep veto power — the drag-back loop still works before expiry, and everything executed is visible after. The operator's absence degrades the system to its stated defaults instead of freezing it.
The design discipline this forces is healthy: if you wouldn't let a decision's default auto-execute after 7 days of silence, then either the default is wrong or it isn't really a decision with a default — it's a blocking question, and it should escalate (a louder Things task, an email) rather than expire. Sorting your decisions into those two bins is the real lesson.
How long has each open card been waiting? One line answers it:
find '/Users/tom/Claude/FOR TOM/TTR' -name '*.html' -mtime +7 -exec ls -l {} +A default that can't execute is just a description of your indecision — give defaults a deadline and the system survives your absence.