What got dropped

Every meeting produces promises. "I'll send you that." "We should try X." "Let me look into it." The pipeline captures them as directives and tracks what happened. Two years in, that's 401 of them. The breakdown is the honest part.

• 89 done
• 40 open
• 272 dropped

Two-thirds dropped. That number looks like failure. It isn't. It's the most accurate thing the corpus says about how a group actually works.

---

how a directive gets sorted

The triage rule is simple and unsentimental. A recent item stays open — the moment might still happen. An older item that was a communication — send, share, follow up — gets marked done; those things either happened or stopped mattering, and either way they're closed. An older item that was a build — make, implement, develop — gets marked dropped, because if it didn't happen in the months after it was raised, it didn't happen.

No judgment in it. Age plus verb. The point isn't to grade anyone. It's to stop pretending a two-year-old "we should build X" is still a live commitment.

---

what's actually open

The 40 open directives are recent and concrete. They cluster, this month, around one thing — getting the meeting's own tooling into the group's hands:

• test the transcript plugin against real meeting records
• store transcripts in a shared repository the group can reach
• record an instruction video for the plugin
• research Safe Superintelligence and report back

These are alive because they were raised in the last few weeks and the moment hasn't passed. Next quarter, the ones that turned into communication will close, and the ones that were builds-that-didn't-happen will drop. That's the cycle.

---

the case for dropping

The 272 dropped directives are the corpus's quiet argument against a certain kind of productivity theater. A group that meets every Wednesday for two years generates far more intentions than any team can execute. The failure mode isn't dropping them. It's pretending you didn't — carrying a backlog of dead commitments that makes every status honest-looking and useless.

Letting go is a feature. Most ideas should die. The ones that mattered came back — a good idea doesn't depend on a single directive surviving, it recurs until someone builds it. The corpus has a name for that pattern: recurrence is the signal. A theme that returns without anyone deciding to raise it is worth building on. A directive that drops and never comes back was telling you something too.

So the 272 aren't waste. They're the group being honest about scope — saying many more things than it could do, and keeping a record straight enough to admit it. The achievement isn't the 89 done. It's that the system can tell you what it let go, and isn't ashamed of the number.