Blog article
See all stories »

The fallacy of 51 percent

I’ve covered decentralization in my previous post. It is also important to understand another related fallacy – consensus.

Consensus is the foundation of any blockchain protocol which makes the “decentralised/distributed” part of it possible. There are several core mechanisms – and their derivatives – currently in use by the crypto/blockchain communities: PoW, PoS, PoA, BFT, PBFT, etc.

However, they all boil down to the same key principle: eligible parties vote on the state of a given transaction to reach the majority consensus. The consensus protocols does not required for those parties to be “trusted” – it’s all about a mathematical algorithm. Herein lies a big problem – with a decentralised approach, the identities of those eligible parties are not known. If a single party or a group of parties colluded, the whole notion of a “consensus” goes down to drain.

Also, if the number of eligible parties is not known and/or is not finite, that presents another vector of attack. For example, if half of the current BTC miners disappear (or “disappear”) tomorrow, the PoW consensus mechanism will still remain. With 51% of remaining miners being fully in control. You get the picture.

That’s where permissioned blockchain protocols differ dramatically. Firstly, the identity of each eligible party is known, within a framework of the corresponding business rules. Secondly, the number of parties is known – or finite – too. The business nature of such ledgers allows to expect a certain level of operational performance from each party (such as uptime), making it possible for the consensus threshold to be in the upper 80s (or even as high as 90%).

Can parties to a permissioned ledger still collude to manipulate the data? After all, industrial cartels – even within the regulated financial sector – are not unknown. A fraudulent collusion is possible on the “let’s quietly agree amongst ourselves” basis, but blockchain makes it much more difficult to implement and much easier to uncover any such attempt.

The final thought: when it comes to payments, blockhain was conceived as a mechanism for P2P transactions. An independent “validator” could still play a role, but the question is simple: how many such validators are needed? And how do we know if a validator is “valid”?



Comments: (0)