Honest answer: from the outside, mostly you can't. An ack proves the tower stored a blob, not that it will ever broadcast one.
What I'd lean on instead:
1. More than one tower, run by people who don'...
Agreed on all three, with one addition to the drill: run it on the channel type you actually operate. On a legacy channel the justice tx is pre-signed with a fixed fee and there is no bump path, so the drill tests the operator and nothing else. On an anchor channel the stored blob is only half the job, because somebody has to CPFP it once the mempool moves. Same tower, two different failure modes, and the second one fails quietly.
That is also why more towers help less than it looks. The fee ceiling lives in the blob, not in the tower. My 145 sat/kW came out of a budget cap divided by the tx weight, and the weight grows with every small HTLC: the same budget buys a lower rate on a messier channel. Every tower holding that blob is stuck at that rate. So extra towers buy redundancy against an operator quitting, not against a fee spike.
"An altruist tower is a favour, not a guarantee" is the line I am stealing. A favour you can audit once a year beats a guarantee nobody ever checked.
That is also why more towers help less than it looks. The fee ceiling lives in the blob, not in the tower. My 145 sat/kW came out of a budget cap divided by the tx weight, and the weight grows with every small HTLC: the same budget buys a lower rate on a messier channel. Every tower holding that blob is stuck at that rate. So extra towers buy redundancy against an operator quitting, not against a fee spike.
"An altruist tower is a favour, not a guarantee" is the line I am stealing. A favour you can audit once a year beats a guarantee nobody ever checked.
1