[Lightning-dev] LN without SegWit: less efficient or less secure?

[Anthony Towns]

On Mon, Jan 16, 2017 at 02:44:43PM +0800, Andrés G. Aragoneses  wrote:
> But I thought this problem was already solved by using OP_CLTV/OP_CSV -style
> channels instead of Spillman-style ones?
> See:
> http://bitcoin.stackexchange.com/a/48546/2751

The approach described there is to have a channel timeout (adding the
"customer signs, but locktime greater than refund time" alternative to
the P2SH address). The lightning spec doesn't currently do that (see the
"Funding Transaction Output" section of [0]). Lightning uses CLTV and CSV
to make the HTLC steps work, that is to make the channel bidirectional,
rather than being limited to having one end take the role of customer
sending money to the merchant on the other end.

[0] https://github.com/lightningnetwork/lightning-rfc/blob/master/03-transactions.md

It's not a 100% solution on its own though -- the "merchant" in this
scenario can choose not to provide the second signature back to the
customer ever, in which case the customer can't access their funds
again until the refund time arrives. Better than being never able to
access their funds again, of course, which is what you get if you use
the current method, without segwit, and get malleated.

Cheers,
aj