[WITHDRAWN] Alternate script design for LNHANCE-Symmetry

[reardencode]

Hey AJ, thanks for your response!

@ajtowns

• This is missing the <n> CLTV step that makes eltoo work

Yep, left it out of both, so doesn’t change the comparison really.

@ajtowns

• <settlement-sig> <G> CHECKSIG is just a more expensive way of doing <h> CTV; it’s not surprising that changing this to CTV makes it cheaper. I believe you can make it cheaper still by using scriptless scripts to reveal a signature instead of doing the CTV-ish approach.

Yeah, CTV+APO is even more efficient, I’m just comparing an APO-only or CTV+CSFS-only world.

@ajtowns

I don’t think this works at all? If you grind the IPK, the scripts you end up with are something like:

• <n+1> CLTV DROP musig(A,B) CHECKSIG
• INTERNALKEY CTV

The scriptPubKey is then calculated as IPK + Hash(IPK, Hash(Hash(S1), Hash(S2))).

But to spend using the first path, you need the witness stack to contain IPK, Hash(S2), S1, signature. But to know what IPK is, you need to know what the settlement tx corresponding to this update tx is, and to redo the grinding. The ln-symmetry approach to solving that is just to stick it into the annex of the update tx that you’re looking to spend.

Ah, but because the full script content is deterministic, you can compute the settlement hash / IPK by calculating the the taptweak (t) and then IPK=SPK-t*G. By moving the pieces around we solve the data availability problem.

Goddamnit, this doesn’t work at all. My cleverness defeated. You need to know the IPK to calculate the tweak - otherwise the whole taproot construction would be insecure (any script tree could spend any outpoint).

@ajtowns

I think lnhance would be more efficient with an IF than two leafs, namely:

Yeah. That applies equally to both APO-Symmetry and LNHANCE-Symmetry, but in a real implementation it would indeed be advisable to use an IF instead of two tapleaves.

Ohhh!!! It doesn’t apply to both, because the APO settlement leaf is large enough that it makes updates worse. Nice. I’ll update OP.

@ajtowns

• • lnhance: 33B annex, 33B control block, 48B script, 64B signature, 32B ctv hash (210B)

A stray annex in this line