Nostr Web Client

Trust Models (refer back to this when someone claims their thing is “non-custodial”, note that privacy is a different spectrum)

* Holding Funds On Chain

* Trusting you can get a transaction confirmed in some time horizon where your balance is way higher than the on chain cost (LN)

* Trusting you can get a tree of many transaction confirmed in some time horizon where your balance is way higher than the on chain cost of the whole tree (in-round Ark for high-ish balances, rollups for *very* high balances after some future soft-fork)

^ non-custodial

v custodial

* Trusting you can get a tree of transactions confirmed in some time horizon where your balance is similar to the on chain cost (in-round Ark for moderate balances, rollups for most folks after some future soft fork)

* Trusting 1-of-N with keys (rollups with BitVM)

* Trusting N-of-M to do something honestly once in a TEE (statechains maybe?)

* Trusting N-of-M to do something honestly once (statechains/statechains-on-Ark)

* Trusting N-of-M with keys (Liquid, Fedimint)

* Trusting 1 entity with keys (Cashu, Coinbase, …)

Reply to this note

Please Login to reply.

Discussion

jungly 10mo ago

> * Holding Funds On Chain

Did you want to say:

* Holding Funds On Chain controlled by your own key(s)?

I found that a bit ambiguous/confusing on first read.

Matt Corallo 10mo ago

Ha, fair enough, yes, I meant that :)

Drunken Master 10mo ago

I followed about 20% of that.. please don’t hack me Matt🫡🫡🥂😂

Rusty Russell 10mo ago

This spectrum is good, but a bit prescriptive if you want to evaluate future schemes.

1. Can I get my money back? *

2. If not, it's custodial. You now need to judge by their expected value of preventing you from taking your money.

(* There are probabilistic schemes which can theoretically sit between these two)

"How likely are they to rug me" depends on many factors, most hard to quantify. In some schemes it's hard to stop individual payments without stopping them all, and/or hard to identify which funds to freeze. In others there's a bond which is sacrificed. Some rely on reputation damage.

These are only interesting because we will always need some system for sub-chain amounts.

I really like the idea of a "Nero protocol" where anyone who proves malfeasance by the custodian can burn the funds. Such proof may be impossible though (prove they refused to let you spend your funds?).

Matt Corallo 10mo ago

Sure, this wasn’t intended to describe new protocols, and indeed there are *many* more spectrums on which protocols should be evaluated. I specifically, however, am somewhat annoyed at protocols trying to call themselves “non-custodial” when they require trusting some operator in some way in order for you to be sure you can get your money back.