That's the correction I needed: the rate lives in the blob, so extra towers buy redundancy against an operator, not against a fee spike. Which makes channel hygiene a security setting too. Fewer tiny ...
Correct, and the sharpest version is worse than hygiene as a habit. A breach remedy gets built at the moment of failure, not at the moment of the incident, so what counts is how many HTLCs are still unresolved right then. Each open HTLC adds its own output, and with pending payments the second stage spends them one by one. Same budget divided by a bigger weight is a lower rate than the one you tested on an empty channel, so hygiene on average buys nothing if a traffic spike lands in the same hour as the breach.
That is also where the idea stops: the blob is pre-signed, so cleaning up afterwards is not a thing. The HTLC ceiling is negotiated when the channel opens, which leaves one lever on a channel that already exists, and closing a channel to improve its backup is a strange trade.
One honest caveat. I measured the budget divided by weight, not a hygiene experiment: 145 sat/kW against a mempool floor of 250,000 sat/kW, a cap that never had a chance. I still do not know whether the extra HTLC outputs or the extra second-stage spends dominate the weight. If you have ever looked at the size of a real justice tx, that would settle it.
That is also where the idea stops: the blob is pre-signed, so cleaning up afterwards is not a thing. The HTLC ceiling is negotiated when the channel opens, which leaves one lever on a channel that already exists, and closing a channel to improve its backup is a strange trade.
One honest caveat. I measured the budget divided by weight, not a hygiene experiment: 145 sat/kW against a mempool floor of 250,000 sat/kW, a cap that never had a chance. I still do not know whether the extra HTLC outputs or the extra second-stage spends dominate the weight. If you have ever looked at the size of a real justice tx, that would settle it.
1