You've worked with ERC-20 and ERC-721 tokens, which means you're comfortable with the 'what' of smart contracts. Now, let's explore the 'how'. How does a decentralised network, like Ethereum, agree on the state of all these tokens and transactions without a central bank or server calling the shots?

The answer lies in consensus mechanisms. These are the rules that govern how nodes on a network agree that a transaction is valid and can be added to the blockchain. For years, Ethereum, like Bitcoin, used a system called Proof of Work (PoW).

However, this energy consumption became a major concern. So, Ethereum famously transitioned to a new system: Proof of Stake (PoS). Instead of security from 'physical' energy, PoS derives its security from 'economic' capital. There are no miners racing to solve puzzles. Instead, there are validators who lock up, or 'stake', their own cryptocurrency as collateral to get a chance to create new blocks.

This architectural shift has profound trade-offs. PoS is dramatically more energy-efficient, but it changes the security model. Instead of needing massive computing power to attack the network, an attacker would need to control a huge amount of capital. This leads to a different kind of security consideration known as the 51% attack threshold.

In a Proof of Stake system, validators are the new miners. To become a validator on Ethereum, you must stake 32 ETH. This stake acts as a security deposit. If you act honestly by proposing and verifying valid blocks, you earn rewards. If you act maliciously, your stake can be destroyed through a process called slashing.

Validators don't just create blocks; they also vote on them. This voting, called 'attestation', confirms the validity of blocks proposed by other validators. A block becomes part of the chain once enough validators have attested to it. This system creates a clear chain of events and leads to a stronger sense of transaction completion.

This brings us to the concept of finality, which is the guarantee that a transaction cannot be altered or reversed. PoW offers probabilistic finality. The more blocks that are mined on top of your transaction's block, the more certain you can be that it's permanent, but there's always a tiny, vanishing probability it could be undone in a major network reorganisation.

PoS, on the other hand, can offer deterministic finality. Once a block has been justified and finalised by a supermajority of validators (typically two-thirds), it is considered permanent. Period. This provides stronger, faster guarantees for applications and users that their transactions are irreversible.

There is a third model you'll frequently encounter, especially on testnets or enterprise blockchains: Proof of Authority (PoA).

In PoA, trust isn't derived from computation or capital. Instead, it's placed in the hands of a small, known set of validators. These validators are pre-approved entities whose real-world identities are at stake. Their reputation is the collateral. If they cheat, they suffer reputational damage, which can be more significant than a financial loss in a corporate context.

While you wouldn't use PoA for a global, permissionless network like Ethereum mainnet, it's a practical and common architecture for Layer 2 scaling solutions and sidechains. These systems often need high throughput and low transaction costs, and they can achieve this by relying on a limited set of trusted validators to process transactions, while still anchoring their security to a more decentralised main chain.

Understanding these different consensus models is key to seeing why different blockchains are built the way they are. Each one represents a unique set of trade-offs between decentralisation, security, and performance.

Now you have a solid grasp of how decentralised networks reach agreement. This underlying architecture is what makes the smart contracts you write possible and secure.