/ratchet:auction — friction auction

Not all obstacles deserve your attention. This command runs a sealed auction where blockers bid for your effort, and the highest-leverage one wins. It exists to stop you from doing work that is interesting but does not advance the target.

Step 0 — Load state

ratchet status

Procedure

1. Enumerate the real blockers between the current state and the locked target's proof-of-done. Real means: if unsolved, the target does not ship. Exclude anything that is merely uncomfortable or merely interesting.
2. Score each blocker on four 1–10 axes:
   • Leverage — how much unblocking it frees downstream (10 = unblocks everything).
   • Certainty — how sure you are it is actually the blocker (10 = proven).
   • Time-to-unblock — how fast it can be cleared (10 = fast, 1 = slow grind).
   • Risk — cost of ignoring it (10 = the whole thing fails if left).

3. Compute priority with the CLI so the math is reproducible:

   ratchet score friction '[{"name":"...","leverage":9,"certainty":8,"speed":6,"risk":9}, ...]'

   Priority = Leverage × Certainty × Time-to-unblock × Risk. Highest priority wins.

4. Defend the winner. State why it beats the most tempting alternative — usually the one that is more fun, more visible, or already familiar.
5. Name the artifact that removes the winning bottleneck. That artifact is the target of the next /ratchet:build.

Output contract

FRICTION TABLE: <the ranked table from the CLI>
WINNING BOTTLENECK: <name + priority>
BEATS: <tempting alternative> because <load-bearing reason>
REQUIRED ARTIFACT: <what to build to clear it>

Serialize

ratchet state set bottleneck "<winning blocker>"
ratchet state set phase auction

Next: /ratchet:cut to make sure the bottleneck is real before you spend on it, then /ratchet:build.