Consensus mechanism
A consensus mechanism is the rule by which a network with no centre decides which version of history is correct. It sets who writes the next block and what keeps a participant honest.
The problem it solves
Thousands of nodes receive transactions in different orders and with different delays. They must arrive at one version of events without trusting each other and without an arbiter. That is the consensus problem.
The difficulty is that some participants may lie deliberately. The mechanism has to work even while a minority actively attacks, and to make attacking cost more than it returns.
Different mechanisms stake different resources. Proof-of-Work makes writing expensive, Proof-of-Stake makes breaking the rules expensive, and permissioned networks simply vote among known parties.
How they differ
What an attacker actually loses under each approach.
| Proof-of-Work | electricity and hardware |
| Proof-of-Stake | the locked stake |
| Permissioned networks | reputation and access |
Hence the differing properties: the first is resilient but energy-hungry, the second efficient but prone to concentration among large holders, the third fast but reliant on trusting the participant list.
What to keep in mind
- Speed is bought with trustNetworks doing a thousand transactions a second usually have fewer independent validators. That is a trade-off, not superiority.
- Finality variesIn some networks a transaction is final within seconds; in others the chance of reversal merely shrinks with each block.
- Nodes decide, not hashpowerA miner proposes a block, but nodes accept it. Rules change only with their agreement.