SELF-CUSTODY FOR BILLIONS NEEDS TWO SOFT FORKS FIRST

A paper landed on Delving Bitcoin at 4:38 Tuesday afternoon. It is called Depots, and it claims to give billions of people self-custodial bitcoin. It is about forty-eight thousand characters long. As of this writing it has seven views, no likes, and not one reply from anybody but the author.

Read the first paragraph anyway, because it is the most honest sentence written about self-custody this year.

The author sets three requirements for any protocol that lets billions of people hold their own keys. A tiny on-chain footprint, on the order of one or two vbytes per user per year. Participation by people who do not own enough bitcoin to pay an on-chain fee. And no more off-chain transactions than can be settled on-chain before theft occurs.

Then, in his own words: no trust-free scaling protocol meets these requirements, so it is widely believed that most users will have to rely on custodial, trust-based solutions such as exchanges.

That is a proposal author describing the field he is trying to fix, and it is the number that operators should sit with. Not your keys, not your coins is a moral claim. One or two vbytes per user per year is an engineering claim, and right now nothing clears it.

What he proposes is a depot. An operator puts one funding transaction with a single Taproot output on-chain. That output backs off-chain Lightning channels for thousands or millions of users, with the depot parameters and a merkle-sum tree of who owns what committed into the key. Channels are probabilistic. Users guess, guesses are committed with salts so the operator cannot know them in advance, and after expiry the operator reveals the target.

Security does not come from cryptography making theft impossible. It comes from what the paper calls griefer penalization. Either side can make the other lose funds, but only by giving up a proportional expected amount of their own. In his framing, someone who only pairs with self-interested parties never gets griefed.

That is an economic guarantee, not a mathematical one, and the paper says so. Its own list of limitations runs six items: reliance on griefer penalization, reliance on expected values, potential to delay access to users’ funds, bounds on large atomic sends and receives, and two separate kinds of capital inefficiency.

And it does not work on bitcoin as it exists. Depots require CTV and CSFS, two consensus changes that have been argued over for years and are not activated. The author lists three more operators that would make it far more efficient and are also not there.

So the state of the sovereignty question, on the day the Senate spent its afternoon on something else, is this. Holding your own keys at global scale is a block space problem before it is a political one. The people working on it are not pretending otherwise. Every route anybody has drawn runs through a consensus change bitcoin has not agreed to make, and the argument about those changes is the argument that actually decides whether most people ever hold their own coins.

Whether this particular design survives contact with the people who read that forum is unknown, because none of them have read it yet.