I’ve seen claims like @BabylonLabs_io ’s again: “No need to trust off-chain = zero cost.” I really can’t hold it in anymore, so let me talk for a bit.
When many people bring up “no need to trust” cross-chain, they usually assume it means “zero fees, zero maintenance.” After checking what the off-chain costs actually are, I found it’s not at all the case. Generating a single circuit with a single core takes about 20 minutes—yes, it can be parallelized, but each side still has to store the counterparty’s circuit. One circuit is 43GB. That storage volume isn’t small. Even though when you do the math the cloud storage cost is roughly $1/month and sounds cheap, the premise is that someone keeps maintaining this entire infrastructure. #baby
What really helped me understand the logic is this line: large borrowers may be more willing to generate and store the circuits themselves, keeping the “no need to trust” property fully intact; small users are likely to delegate the work to professional operators. Here’s a key guarantee: operators can’t steal BTC, because spending still requires the borrower’s co-signature. Operators are just doing the heavy lifting—they’re not custodians. Also, circuits can be reused, not regenerated for every deposit. The losing party is automatically eliminated by the system (if the operator loses the case, they are removed and lose the future right to withdraw; if Larry loses, he loses the right to challenge). “Pay once, account once”—after amortization, the actual cost is much lower than you’d think.
What’s smart about this design is that it cleanly separates “who should bear the off-chain maintenance cost” from “who controls asset custody.” Delegating maintenance doesn’t mean transferring trust. That’s the point I hadn’t held together before—I assumed that as soon as a third party is involved, it inevitably introduces trust risks.
Personally, I lean toward having small amounts of money go through operator-managed custody to save hassle. But the specific mechanism by which an operator is eliminated after losing a case—who decides, and how long it takes to take effect—those on-chain execution details are something I didn’t find clearly. It’s probably still being designed. If off-chain maintenance can be outsourced but trust can’t, would you still care about the question “who is helping you store the circuits”? #baby $BABY
When many people bring up “no need to trust” cross-chain, they usually assume it means “zero fees, zero maintenance.” After checking what the off-chain costs actually are, I found it’s not at all the case. Generating a single circuit with a single core takes about 20 minutes—yes, it can be parallelized, but each side still has to store the counterparty’s circuit. One circuit is 43GB. That storage volume isn’t small. Even though when you do the math the cloud storage cost is roughly $1/month and sounds cheap, the premise is that someone keeps maintaining this entire infrastructure. #baby
What really helped me understand the logic is this line: large borrowers may be more willing to generate and store the circuits themselves, keeping the “no need to trust” property fully intact; small users are likely to delegate the work to professional operators. Here’s a key guarantee: operators can’t steal BTC, because spending still requires the borrower’s co-signature. Operators are just doing the heavy lifting—they’re not custodians. Also, circuits can be reused, not regenerated for every deposit. The losing party is automatically eliminated by the system (if the operator loses the case, they are removed and lose the future right to withdraw; if Larry loses, he loses the right to challenge). “Pay once, account once”—after amortization, the actual cost is much lower than you’d think.
What’s smart about this design is that it cleanly separates “who should bear the off-chain maintenance cost” from “who controls asset custody.” Delegating maintenance doesn’t mean transferring trust. That’s the point I hadn’t held together before—I assumed that as soon as a third party is involved, it inevitably introduces trust risks.
Personally, I lean toward having small amounts of money go through operator-managed custody to save hassle. But the specific mechanism by which an operator is eliminated after losing a case—who decides, and how long it takes to take effect—those on-chain execution details are something I didn’t find clearly. It’s probably still being designed. If off-chain maintenance can be outsourced but trust can’t, would you still care about the question “who is helping you store the circuits”? #baby $BABY