Hi ZmnSCPxj,

>potentially both Alice and Bob know all the secrets on the LTC side and end up competing over it

That's exactly right.

>Bob can thus give a copy of the revoke tx with signature directly to its favorite miner, forcing Alice to take 3 transactions

Note that the timelock on the revoke tx is longer than the timelock on refund tx #1. The idea is that Alice aborts the protocol by publishing refund tx #1 if the protocol hasn't reached step 4 in the svg by the time it becomes valid. This should entirely mitigate the issue you're describing.

>adding two CPFP outputs (one for each participant)

There seems to be a situation where RBF can be disabled by the other party, but I'm not sure I see it... Why would a single output spendable by either key be insufficient?

>We could use `SIGHASH_SINGLE | SIGHASH_ANYONECANPAY` as well

Allowing others to add inputs/outputs would introduce malleability. Refund tx #2 and the timeout tx would become invalid.

>Bob cannot safely perform step 2 before getting both signatures for the revoke tx

That's right, as you guessed, he does receive a copy of the signed revoke tx at protocol start.

>>alternatively Bob can just spend before the timelock expires.
>This seems to be the safest alternative

I agree not giving Alice time to publish the revoke tx is safest, but one does not preclude the other. The revoke tx is on an absolute timelock, so spending it before that time means you don't have anything to worry about, and spending it later means you'll have to be online and keep an eye out. If staying online is not a problem, then fee wise that seems preferable. As long as less than half of all valid (i.e. the timelock was reached) revoke transactions get broadcast, you'll be saving on fees.

Cheers,
Ruben

On Wed, May 13, 2020 at 11:57 AM Ruben Somsen <rsomsen@gmail.com> wrote:
Hi Chris,

Thanks for taking a look :)

>it also improves privacy because the coins could stay unspend for a long time, potentially indefinitely

Excellent point. The pre-swap setup transactions would still be subject to timing/amount analysis, but it's clearly a lot less problematic than the traditional 4 tx swap. And Payswap may be able to mitigate the amount analysis.

>Using relative timelocks and private key handover for old-style coinswaps would give us the same two-transaction effect

I agree, Lloyd pointed out the same thing. One thing to add is that such a setup would result in four on-chain transactions if the protocol is aborted, due to the need to invalidate the refund transaction.

>the idea of private key handover was mentioned as early as 2016 in the original Lightning Network paper

Interesting! Thanks for pointing that out.

Cheers,
Ruben

On Wed, May 13, 2020 at 10:39 AM ZmnSCPxj <ZmnSCPxj@protonmail.com> wrote:
Good morning Ruben,

> >If the shortened refund transaction exists (labeled "refund transaction #1" in the SVG) then the same issue still occurs 
>
> Yes, but there is one crucial difference: at that point in the protocol (Bob has the success transaction and then stops cooperating) Alice and Bob both had the opportunity not to take that path. Bob could have sent the success transaction, and Alice could have waited and sent the revoke transaction. They would essentially be "colluding" to fail.

Okay, so the concern is basically, that Bob misses the deadline, then Alice feels obligated to reclaim the funds.
In your proposal, the tx competition is between the secret-revealing success TX and the non-secret-revealing revoke tx.
Whereas in my counterproposal, under the same conditions, the tx competition is between the secret-revealing success tx and the secret-revealing backout tx, and both transactions becoming visible on P2P network means potentially both Alice and Bob know all the secrets on the LTC side and end up competing over it, RBFing each other until the entire fund goes to miners.


> >Without the refund#1 in your proposal, Bob refusing cooperation after Alice puts the BTC into lock for 3 days and 2 further onchain transactions
>
> I'm not sure if I correctly understood what you're saying, but it's as follows:
>
> Refund #1 can only safely be used before the signed success tx is given to Bob. The cost to Alice at this point if Bob aborts is two on-chain transactions while Bob hasn't put anything on-chain yet.
>
> Refund #2 needs to be used after Bob receives the signed success tx. The cost to Alice is now three transactions, but Bob also went-on-chain by this point, so causing this wasn't costless to Bob and is thus a similar failure mode.

I think it is not accurate that Bob is already on-chain before Alice can be forced to use 3 transactions to fail.

The revoke tx signatures are shared at step 0 of your protocol description.
Thus Bob has a copy of the revoke tx that is valid once Alice signs and confirms the funding transaction.
Bob can thus give a copy of the revoke tx with signature directly to its favorite miner, forcing Alice to take 3 transactions to back out of the swap.

Since Bob gave the tx directly to its favorite miner (TANSTAAGM: "There ain't no such thing as a global mempool"), Alice will only know about this event when the revoke tx is confirmed once, at which point it is very difficult to reverse, even if Alice has a refund#1 tx prepared.

Bob can do this before step 2 in your protocol description, meaning before Bob locks up any funds, so Bob can do this for free, and will even use funds going back to Alice to pay for confirmation of the revoke tx.
Because Bob can do this for free, there is no disincentive for trolling Bobs to exist whose sole life goal is to just grief possible Alices.

This can be slightly mitigated by adding two CPFP outputs (one for each participant) and using the minimum relayable feerate for the revoke tx so that Bob is forced to bring its own fees in order to incentivize miners.
This is similar to the "bring your own fees" currently proposed for Lightning, but note the recent hand-wringing about the various problems this adds to mempools and CPFP and RBF rules and etc etc: https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2020-April/017757.html

We could use `SIGHASH_SINGLE | SIGHASH_ANYONECANPAY` as well for a bring-your-own-fees, but that is not `SIGHASH_ALL` and thus marks the transaction graph as special.
And forcing bring-your-own-fees means neither Alice nor Bob can swap all their funds in a single operation, they have to keep a reserve.


Bob cannot safely perform step 2 before getting both signatures for the revoke tx, as without Bob having access to the rveoke tx, if Bob locks up LTC, Alice can stop responding and lock both their funds indefinitely with Bob not having any way to recover its funds, which a rich Alice can use to completely lock out an impoverished Bob.
But if Bob is given both signatures for the revoke tx before step 2, then Bob can send the revoke tx to its favorite miner, forcing Alice to take 3 transactions to back out, before Bob locks any funds in LTC side.

>
> I also agree with your observation that alternatively Bob can just spend before the timelock expires.

This seems to be the safest alternative; in my context, where Bob is a CoinSwap server/maker, Bob can wait a short while for new clients/takers, and if no new clients arrive, spend.
Bob can run multiple servers, each of which are given the completed success transaction, and the servers can check that if the timeout is near, to spam the Bitcoin P2P network with the completed success transactions.
(these servers need not even run fullnodes, they could just periodically poll a number of blockchain explorers and electrum servers, and when the blockheight approaches, attempt broadcast; if the "main" server that accepts clients/takers has already spent the TXO the broadcast of the completed success tx is simply rejected by the Bitcoin P2P network; if the timeout is based on sidereal time then the backup servers only need to be running NTP)



Regards,
ZmnSCPxj