The issue of settlement finality has recently surfaced in the context of blockchain technology. Tim Swanson of R3CEV and Vitalik Buterin of Ethereum have both dug deep on the issue in the context of permissionless vs permissioned blockchains.
A transaction achieves settlement finality when it has completed, and there is no chance of it reverting or changing state again. This is of particular importance in financial systems, because a buyer of an asset must know, definitively, that the asset ownership has transferred, and a person being paid must be sure the funds will not be reversed or shifted elsewhere.
The take away conclusion of the commentaries is that public, permissionless blockchains cannot guarantee settlement finality.
Vitalik Buterin describes Proof of Work (PoW) issues this way. “Technically, a proof of work blockchain never allows a transaction to truly be “finalized”; for any given block, there is always the possibility that someone will create a longer chain that starts from a block before that block and does not include that block.”
Proof of Stake (PoS) blockchains use different logic, but can also not ultimately guarantee settlement finality.
The protection both PoW and PoS rely on is economic incentive. To attack a transaction or a set of transactions will result in great expense to an attacker. Therefore the reward has to outweigh the cost of the attack. For lower value transactions it is reasonable to assume an attacker will be de-incentivized by the cost of an attack. However, for high value transactions an attack may result in a greater reward than the penalty.
Permissioned blockchains however, can achieve settlement finality. The participants, who are known and identified, and have a commercial relationship – either directly or by virtue of being authorized participants on the private network that is the blockchain – work within a legal framework outside the blockchain.
Some critics of permissioned blockchains point to instances where large financial institutions have acted in an untrustworthy manner, such as the collusion to manipulate the Libor rates. They argue that this means a network of such organizations cannot be trusted without the public oversight granted by the permissionless blockchains. They forget that on a regulated system, which such a permissioned blockchain must be, there are regulators that serve the public purpose. Any transaction that has any measure of systemic risk requires oversight. A regulator would therefore be privy to the process of settlement finality, and it would be impossible for any attack or manipulation to take place without alerting a participating regulator.
Our conclusion is therefore, that a well-designed, permissoned blockchain is the only solution for large scale, regulated systems that require settlement finality.