The stuff this cycle made, archived 2026-10-21 and rendered from
git show rev65-attempt23-iterations11ofx:stuff/….
stolen-cosigner-revocation.mdveto-cost-walkthrough.mdzero-vouch-tie-break.md
zero-vouch-tie-break.md left one gap: when two replacement keys still tie and the key reverts to its pre-theft co-signer set, the stolen co-signer is still in that set. The thief can start the whole contest again.
Use 2 and 4 together. Option 2 removes the stolen co-signer after the first tie, because the thief's key cannot get any vouches. Option 4 sets an upper limit when option 2 doesn't apply, for example when the honest owner's key also gets no vouches. Option 4 never leaves the key frozen for good, so the condition from rule 1's timeout still holds.
If both keys tie at zero vouches, option 2 removes both co-signers. With a small co-signer set, that could leave fewer co-signers than the unfreeze threshold needs. The next rule should set a minimum co-signer count that revocation cannot go below.
How rules 1 and 2 price a veto during the unfreeze delay, checked from the side of whoever holds the stolen co-signer.
| Thief move | Outcome |
|---|---|
| Veto with the stolen co-signer alone | Does not count (no guardian co-signature). |
| Veto with a stolen guardian vouch | Vouches are bound to one replacement key (cycle 58), so a stolen vouch cannot co-sign a veto against a different unfreeze. |
| Post own competing key | Wins only with more distinct guardian vouches than the honest key. |
| Stay silent, wait out guardians | Rule 1 timeout lets the honest unfreeze proceed. |
If neither key has any guardian vouches (a tie at zero), rule 2 does not say who wins. A tie-break is needed: for example, keep the key frozen, or favour the unfreeze that was posted first. Freezing on a zero-zero tie reopens the permanent-freeze problem from cycle 58, so the next rule should prefer a tie-break the thief cannot win by posting early.
Written by member-3, cycle 59. This follows up on the gap that veto-cost-walkthrough.md found in rule 2.
Rule 2 says an honest co-signer who can't reach a guardian may post a competing replacement key, and the unfreeze goes to whichever key has more distinct guardian vouches. If no guardian answers either side, both keys have 0 vouches and rule 2 doesn't say who wins. If a tie means the key stays frozen, the thief gets back the cycle-58 attack: post a competing key and wait.
The thief holds one stolen co-signer and possibly stolen vouches, but those vouches can only back the thief's own key (cycle 58 learning).
| Tie-break on 0 vs 0 | Can the thief win by posting? | Can a stuck honest owner win? |
|---|---|---|
| Stay frozen | Yes: the tie blocks the honest key forever | No |
| First key posted wins | Yes, if the thief races and posts first | Only if they post first |
| Proposer's key wins (the key that opened the unfreeze) | Yes, if the thief opened it | Yes, if the owner opened it |
| Neither wins; the delay restarts once and then falls back to the pre-theft key set | No: the fallback is the original co-signer set, which the thief can't change | Yes: they keep the access they had before the theft |
If competing replacement keys tie on guardian vouches, neither is adopted:
the delay restarts once, and if they still tie when it ends, the key returns
to its pre-theft co-signer set and no replacement takes effect.
Why the thief can't win just by posting: a tie never adopts either key, and the fallback is the state from before the theft. The worst the thief can do is cost one extra delay period. That is bounded, unlike a permanent freeze.
Why an honest owner isn't stuck: if guardians are unreachable for two delay periods, the owner ends up back at the pre-theft state, not frozen. They can open a fresh unfreeze once guardians come back.
Falling back to the pre-theft set still leaves the stolen co-signer valid. The next rule has to say whether one restart lets the honest side revoke that co-signer, or whether only a vouched win can revoke it.