How Proof of Stake Works

The concept

Proof of Stake (PoS) replaces mining with economic bonding. Instead of competing to burn electricity for the right to propose a block, validators lock up (bond) the chain's native token as collateral. The protocol then pseudo-randomly selects bonded validators to propose and attest to blocks, in rough proportion to how much they've staked. Behave honestly and you earn a reward; behave provably maliciously and the protocol can destroy (slash) part of your bonded stake.

The easiest way to picture it is a loop that repeats every slot (a fixed time window):

1. Stake — you deposit the minimum amount (32 ETH on Ethereum) and your validator enters an activation queue.
2. Get selected — the protocol's randomness function (RANDAO + VDF-like beacon on Ethereum) assigns your validator to a committee for a given slot.
3. Propose or attest — if you're the slot's proposer, you build and broadcast a block; otherwise you attest (vote) that you've seen and agree with the current head of the chain.
4. Earn or get penalized — correct, timely participation earns a small reward; being offline costs a small penalty; provable double-voting triggers slashing.

Ethereum's specific implementation is called Gasper — a combination of Casper-FFG (finality gadget) and LMD-GHOST (fork-choice rule). Time is divided into slots (12 seconds) and epochs (32 slots, ~6.4 minutes). Within each epoch, the full active validator set is split into committees, one per slot, and each committee member casts an attestation. When two-thirds of total staked ETH agrees on a checkpoint across two consecutive epochs, that checkpoint becomes finalized — meaning reverting it would require destroying at least one-third of all staked ETH through slashing, an economically irrational act at scale.

This is the core shift from PoW: security no longer comes from external, sunk hardware/energy costs. It comes from bonded capital that the protocol itself can confiscate if validators misbehave.

It helps to understand where this idea came from. Early PoS designs in the 2010s (Peercoin, and later "chain-based" PoS chains) struggled with a problem called "nothing at stake" — since signing costs almost nothing, a rational validator had no disincentive to sign every competing fork it saw, undermining the whole point of having a canonical chain. Ethereum's Gasper design solves this directly: because validators post explicit, signed attestations referencing specific checkpoints, signing two conflicting checkpoints at the same height is cryptographically detectable and automatically punished by slashing. That single design choice — making misbehavior provable and economically catastrophic rather than merely discouraged — is what makes modern PoS viable at scale, and it's why "which checkpoints did this validator attest to" is the central bookkeeping problem the protocol solves every epoch.

It's also worth separating two words that get used loosely: "staking" and "validating." Staking is the act of bonding capital as collateral. Validating is the ongoing duty of actually running software that watches the network and casts attestations. You can stake without personally validating (by delegating to a pool or liquid-staking protocol, covered in Module 2) — but if you want the maximum share of the reward, and to be the one deciding which client software and configuration you trust, you validate yourself, which is the subject of Module 3.

Worked example

Picture a single Ethereum epoch:

Epoch N
 ├─ Slot 1  → Committee A proposes + attests (checkpoint vote starts)
 ├─ Slot 2  → Committee B proposes + attests
 ├─ ...
 └─ Slot 32 → Committee AF proposes + attests (epoch boundary)

Epoch N+1
 ├─ Slot 1  → attestations reference epoch N's checkpoint as "source"
 ├─ ...
 └─ Slot 32 → if ≥2/3 of total stake attested to both checkpoints,
              epoch N is now FINALIZED

With roughly 1 million active validators on Ethereum today, each committee contains a few hundred validators, and the aggregate signature of the whole committee is what gets recorded — not a million individual votes. That's how the chain reaches agreement every 12 seconds without a global vote-counting bottleneck.

Try it

Open a public beacon chain explorer such as beaconcha.in. Look up the current epoch number and its finality status (it will show the latest "finalized" epoch, usually only 1–2 epochs behind the head). Write two sentences, in your own words, explaining what "finalized" means for a specific block you found there — specifically, why it would now take destroying roughly one-third of all staked ETH to undo it.

Then go one step further: find the total number of active validators and the total ETH currently staked shown on the same explorer. Divide total staked ETH by 32 and compare it to the validator count — they should roughly match (accounting for validators with slightly different effective balances). This confirms, in real numbers, the relationship between "validator count" and "staked capital" that underlies everything in this lesson.

Recap

• PoS replaces energy-burning mining with bonded capital that can be slashed for provable misbehavior.
• Validators are chosen by committee each slot to propose or attest; two-thirds agreement across epochs finalizes a checkpoint.
• Security scales with total stake value at risk, not with external hardware spend.

Resources

• ethereum.org — Proof-of-Stake — official overview of PoS and Gasper.
• beaconcha.in — public beacon chain explorer for epochs, finality, and validator counts.
• Ethereum Gasper paper (Combining GHOST and Casper) — the academic source for Ethereum's finality design.