Rubix Insights Research paper

Proof of Pledge: collateral without custody in an unspent-token model

Most staking designs ask a participant to hand over custody of the asset that secures the network. Proof of Pledge does not. A validator commits specific, unspent tokens as collateral while continuing to hold them, and the commitment is checkable by anyone without trusting the validator's own report.

Published
24 July 2026
Written by
Rubix Protocol Research
Reading time
11 min

The custody problem

Staking is usually described as locking value. In practice it is usually transferring value. The participant sends tokens to a contract or an operator, and from that moment their security depends on that counterparty behaving. The asset is gone from their control even though the economic claim remains theirs.

That trade is accepted because it is convenient to implement, not because it is necessary. If the network can verify that a specific token is committed and has not moved, custody does not need to change hands at all.

Pledging an unspent token

Rubix tokens carry their own independent state and their own history. A pledge names specific tokens and records the commitment against them. Because each token's chain is independently verifiable, any participant can confirm three things without asking the pledger: that the tokens exist, that they are unspent, and that they are currently pledged.

A commitment that only the committer can confirm is not collateral. It is a promise.

The pledge is released the same way it was made, as a recorded event with verifiable ordering. There is no window in which the network believes a token is pledged while its holder has already moved it, because moving a pledged token is not a valid state transition.

What this changes for validators

A validator's collateral stays in their own wallet, spendable the moment the pledge is released and not before. The operational burden of running a validator drops, because there is no custody transfer to arrange, insure or unwind.

It also changes the failure mode. When collateral is custodied, a compromise of the custodian is a compromise of the network's security budget. When it is pledged in place, the blast radius of any single compromise is one participant.

Open questions

Pledge concentration remains an open area: nothing in the mechanism prevents a single identity from pledging across many validators, and the network's security assumptions should be stated in terms of independent pledgers rather than pledge count. That analysis is ongoing.