By - Gas measures computation. The gas price is what you pay per unit, and it is set by an auction, not by the network.
- Ethereum splits the fee into a base fee that is burned and a priority tip that goes to the validator.
- Spikes happen because block space is fixed while demand is not.
Gas fees are the most misunderstood cost in crypto, largely because the name suggests a service charge. It is not one. Gas is the unit of account for computation, and what you pay is the outcome of a continuous auction for a fixed resource.
Understanding it explains why the same transaction can cost two dollars in the morning and sixty in the afternoon, with nothing about the transaction having changed.
What is gas actually measuring?
Work. Every operation a chain performs has a fixed cost in gas units: adding two numbers, writing to storage, calling another contract. A simple transfer consumes a set amount. A complex swap touching several contracts consumes far more.
Gas units are a property of the operation and do not change with market conditions. What changes is the price per unit, which is denominated in the chain’s native asset. The full opcode cost table sits in Ethereum developer documentation.
Why is it an auction?
Because a block holds a limited amount of gas, and more people want in than fit.
Validators build blocks by choosing transactions from a waiting pool. They are economically rational, so they take the ones paying most per unit of gas. If you bid below what others are paying, you wait. This is the same mechanism that governs settlement on proof of work versus proof of stake networks generally, whichever consensus model is in use.
What is the base fee and the tip?
Ethereum splits the payment. The base fee is calculated by the protocol from how full recent blocks were, and it is burned, permanently removed from supply. The priority fee, or tip, goes to the validator as the incentive to include you.
The base fee adjusts automatically, rising when blocks run full and falling when they empty. That makes fees more predictable than a pure auction, but it does not make them cheap when demand is high. It just makes the increase smoother.
Why do fees spike so sharply?
Because supply is fixed and demand is not.
A popular mint, a large liquidation cascade, or a token launch pulls thousands of people into the same blocks. They bid against each other. The base fee climbs, tips climb faster, and the cost of an ordinary transfer rises even though the transfer itself is unchanged.
Failed transactions still cost gas. The network performed the computation and discovered the transaction could not complete. You pay for the work, not the outcome, which surprises people the first time a swap reverts.
How do you actually pay less?
Time it, or move off the main chain.
Fees follow predictable daily and weekly rhythms tied to when the largest markets are awake. Weekends and off-peak hours are consistently cheaper. Most wallets show a live estimate, and waiting an hour is frequently the difference between five dollars and thirty.
For anything routine, layer twos exist precisely to solve this, batching many transactions into a single settlement on the main chain. The trade is a different trust model, similar in shape to the one discussed in how crypto bridges work. And if you are moving value between chains to chase lower fees, the bridge itself becomes the risk you are taking.
What happens when a transaction runs out of gas?
It reverts, and you still pay.
Every transaction carries a gas limit, the maximum units you authorise it to consume. If execution exceeds that limit, the chain undoes every state change and keeps the fee for the work already performed. Nothing moves, and the money is gone.
This is why wallets estimate the limit for you and why manual overrides are risky. Set it too low and you pay for a failure. Set it far too high and you have simply authorised a larger maximum, though you are only charged for what was actually used.
Why do different chains charge such different amounts?
Because they make different trade-offs about how much computation each block can hold and how many machines must agree on it.
A chain with fewer validators and larger blocks processes more transactions per second and charges less per transaction. It also has a smaller set of parties who must collude to affect the ledger. A chain with thousands of independent validators and conservative block limits costs more precisely because that redundancy is expensive to reproduce.
Neither answer is correct in isolation. Fees are the price of whatever security model the chain has chosen, which is why comparing them without comparing validator counts tells you very little. Research on these trade-offs is collected in public cryptography literature, and the economics are tracked by bodies including the Bank for International Settlements.
What's your reaction?
[…] conditions, which is the trade-off, and it is usually the right one. This interacts directly with how gas fees work, since a failed transaction still costs a […]