a bitcoin vault needs a waiting room

i sat down with a founder in switzerland building a bitcoin vault, and the line i wrote down wasn't about multisig. it was about time. his words... "in my opinion, just multisignature is not enough. you really need the time delay."

most of us think about self custody in one setting. one wallet, one seed, one device. that's fine for spending. it's a terrible way to keep.

think about your house. you carry a wallet in your pocket and you keep the rest somewhere slower to open. nobody keeps their life savings in the front pocket of their jeans.

his product is the slow place. two of three multisig, so no single key can move anything. then a mandatory waiting room on every outgoing transaction, anywhere from two hours to fifteen days, set by you. the moment a spend starts, you get an alert. a trusted guardian gets a quiet heads up too. and inside that window you can cancel.

and there's a back door that only you hold. a recovery path that uses your own two devices and opens after one year per utxo, written into the bitcoin script itself with checksequenceverify and miniscript. no company sits in that path.

why the obsession with delay?

because of how attacks happen now. he pointed to a recent hardware wallet firmware scare as the thing that pushed him, and his read was simple. a coordinated attack doesn't pick at one wallet. it sweeps thousands of addresses in a tight window. multisig doesn't help you if the attacker already holds two keys. time does. time is the one thing a thief can't fake.

there's a nice property underneath this. a vault delay sits on chain where anyone can see it, so you, or a watchtower working for you, can react before the timelock runs out. native vault opcodes like bip-345 are still only proposed. so builders like him are doing it with the script tools bitcoin already has.

i pushed him on the obvious problem. your co-signature is a new trust dependency. structurally that's the same shape as the things that have failed before. you've added a party to my money.

he didn't dodge it. the company can be forced to stop co-signing. it can't be forced to move your funds, because it never holds enough keys to do that alone. and the recovery path can't be blocked by anyone. worst case, you wait a year and walk out with your own devices.

that's the question i'd ask about every helper you ever add to your setup. not do i trust them. what can they be forced to do, and what can't they do no matter who leans on them.

he's been at it about fifteen months, with something like eight hundred to eight hundred fifty people in the beta. early. but the thinking is right.

his practical advice was as boring as it was good. keep three to nine months of living money liquid. vault the rest. use hardware from more than one vendor so one bug can't take everything. keep devices and backups in different places. and test your backups with a small amount before you trust them with a big one.

notice there was zero talk of price in any of this. that's the tell of someone thinking in decades. you don't build a waiting room for money you plan to flip next month.

single sig is a wallet. a vault is a vault. most people only own the first one and call it the second.

the part i keep coming back to is how unglamorous the whole design is. no new token, no clever trick. just the oldest security idea there is, slowing the thief down, written into money that can't be argued with.

banks figured this out with time locks on vault doors. nobody opens them at midnight, not even the manager. bitcoin lets you build that for yourself, and nobody has to hold the key for you.

more builders like him are on the show if you want to hear how the people writing the script think about keeping.

so if someone got two of your keys tonight... how long would they have to wait before your coins were gone?