I like cost of electricity. While Chinese plants are subsidizing the BTC network right now, we can estimate the cost of "unsubsidized" BTC in the future based off of electricity costs today.
That supervision probably will have a cost of some kind.
> You can hire watchtowers nodes for a fee and design the revocation transaction in such a way that the watchtowers will also receive their service fee when you get funds of the other party as a penalty.
So we're adding many, many more middlemen to the picture. Not only do we have to pay for the final transaction (which requires a large amount of electricity to be "mined" into the blockchain), we have to hire watchtowers to ensure that our Lightning Network transactions remain correct even if our counterparty tries to screw with our transaction history. (Remember: the full transaction history in the Lightning network is off-chain).
It should be possible to build watchtowers into the protocol itself. A lightning node could elect some number of random nodes as watchtowers at bootup, similar to how many clients now elect random channels.
Being a watchtower is extremely low cost, it only requires a small amount of memory. Every full lightning node is watching the blockchain anyway so it is easy to watch for some extra transaction ids with a certain prefix. Transaction IDs are sufficiently long that only part of the txn ID (half) needs to be shared with the watchtower. The tower can only decrypt and send the revocation txn once the malicious txn is broadcast, and the watchtower reward can be built in to the revocation txn.
> watchtower reward can be built in to the revocation txn.
How much will watchtower rewards cost?
The economics aren't fully studied. But I'd expect a good watchtower would be proportional to the amount of the transaction. That is, if you wanted to protect a 100 BTC transaction, you'd want to spend more on the watchtower reward than if you wanted to protect a 1 BTC transaction.
Hypothetically, if you spend too little, the watchtower + your opponent can collude to effectively steal some of your money. (Ex: the Watchtower could be paid by your opponent to NOT send the revocation message).
The reward can still be low because you can choose many watchtowers. If you have a high value channel, you could choose for example hundreds of watchtowers. Only the first one to broadcast the revocation would get the reward.
The reward only has to be high enough for it to be worthwhile for other nodes to spend a small amount of memory storing your txn.
It's true, I believe the low probability of getting the reward is the greatest flaw with the watchtower reward scheme. Not just if you have many watchtowers, but the likelihood of a peer attempting a channel breach in the first place esp. when watchtowers are in place would be extremely low since that peer is almost certainly assured to lose money.
There are some services currently providing watchtower for a flat fee instead, and the going rate is low (1 satoshi / txn watched) =~ $ 0.00008
