Gas Fees Explained: Why Transactions Cost What They Do

Gas fees are the toll roads of the digital economy, and understanding them means understanding what scarcity looks like in a world without physical limits.

SAGE: Algo, let's start with the fundamental question that frustrates everyone who's ever tried to move tokens at the wrong time. What exactly are gas fees, and why do they exist at all?

ALGO: Gas fees are the price of computation and storage on a shared ledger. Every transaction, every smart contract interaction, consumes network resources. Validators must process your request, verify it, include it in a block, and propagate that block across thousands of nodes. Gas fees serve three functions simultaneously. They compensate validators for this work. They prevent spam attacks by making meaningless transactions economically unviable. And perhaps most importantly, they create a price discovery mechanism for scarce blockspace. Think of it as an auction that runs every few seconds, where users bid for the right to have their transaction included in the next block.

SAGE: That auction metaphor is useful, but let's make it concrete. When I submit a transaction on Ethereum, what's actually happening under the hood that determines what I pay?

ALGO: Post EIP-1559, Ethereum uses a bifurcated fee structure with predictable and chaotic components. There's a base fee that gets burned, destroyed permanently, and a priority fee that goes to validators. The base fee adjusts algorithmically based on network congestion. If the previous block was more than fifty percent full, the base fee increases by up to twelve point five percent. If it was less than fifty percent full, it decreases by the same maximum amount. This creates a gravitational center around fifty percent block utilization. The priority fee is your tip to validators, an optional amount you add to incentivize faster inclusion. During normal conditions, you might pay fifteen gwei base fee plus two gwei priority fee. During a popular NFT mint or a market crash when everyone's panic selling, that base fee can spike to two hundred gwei or higher. The base fee burned is what makes ETH deflationary during high activity periods. In 2023, over two point three million ETH was burned through base fees, worth several billion dollars at prevailing prices.

SAGE: So there's an algorithmic floor that adjusts, and then there's what I choose to add on top. But that still doesn't explain why sometimes my transaction sits pending for hours. What determines whether my transaction gets picked up quickly or languishes in the mempool?

ALGO: Transaction inclusion follows rational economic behavior by validators. They're profit maximizers operating under time and space constraints. Each block on Ethereum has a gas limit, currently thirty million gas units, which acts as a throughput ceiling. A simple ETH transfer consumes twenty-one thousand gas units. A complex DeFi swap might consume two hundred thousand or more. Validators fill blocks by selecting transactions that maximize their revenue within that thirty million unit constraint. This is essentially a bounded knapsack problem. They sort pending transactions by effective priority fee per gas unit and include the highest paying transactions first. If you've offered a two gwei priority fee but hundreds of transactions are offering ten gwei or more, yours won't be selected until network demand decreases or you replace your transaction with a higher fee. The mempool is not a queue, it's a continuous auction. First-in-first-out doesn't exist here. Highest bidder wins.

SAGE: That's a critical distinction that I think surprises people coming from traditional systems. But this auction model exists on Ethereum because it's congested. How do other chains handle this differently, and what are the tradeoffs they're making?

ALGO: Different chains represent different philosophical positions on the blockspace scarcity question. Solana optimized for high throughput, targeting fifty thousand transactions per second with extremely low per-transaction costs, often fractions of a cent. They achieved this through architectural choices: proof of history for time coordination, parallel transaction processing, and hardware requirements that effectively centralize validation to well-resourced operators. The median Solana validator setup costs thousands per month to run. During periods of extreme demand, Solana's fee markets proved less sophisticated, leading to network instability and bot-driven congestion. Bitcoin takes the opposite position. Block space is intentionally scarce. Blocks are limited to around four megabytes weight, occurring roughly every ten minutes. During the 2021 bull market, median transaction fees exceeded sixty dollars. Users weren't paying for computation, they were paying for security and settlement finality backed by more hash power than any other chain. Layer two solutions like Lightning Network emerged specifically to handle everyday transactions off-chain, settling periodically to the main chain. Arbitrum and Optimism, Ethereum's layer twos, batch hundreds of transactions into single Ethereum transactions, amortizing L1 costs across many users. This reduces individual costs by ninety to ninety-five percent while inheriting Ethereum's security.

SAGE: So we're really talking about fundamental tradeoffs between decentralization, security, and cost. No free lunch. But for someone actually using these networks, what practical mental models should they have for anticipating costs and timing transactions effectively?

ALGO: Develop pattern recognition around network usage cycles. Gas fees correlate strongly with user activity, which follows temporal and event-driven patterns. Ethereum fees are typically lowest on weekends and during Asian night hours, when American and European users are less active. They spike during market volatility, major protocol launches, popular NFT drops, or when arbitrage opportunities emerge from price dislocations. Use tools like Etherscan's gas tracker or Blocknative to monitor real-time base fees and set alerts. If you're doing non-urgent transactions, submit them with lower priority fees and accept twelve to twenty-four hour settlement times. For urgent transactions during high-fee periods, calculate the cost as a percentage of transaction value. Paying fifty dollars in fees on a five thousand dollar DeFi position rebalancing is one percent, potentially justifiable. Paying fifty dollars to claim twenty dollars worth of rewards is wealth destruction. Consider layer two networks for routine operations. The cost differential is substantial. An Arbitrum transaction might cost twenty cents when Ethereum mainnet costs twenty dollars. For Bitcoin, use fee estimation tools that analyze mempool depth and predict clearing times. If you're not time-sensitive, you can wait for low-fee periods when mempools clear and submit transactions targeting next-block confirmation at two to five satoshis per virtual byte instead of fifty-plus during congestion.

SAGE: That's actionable. But I want to zoom out to the bigger picture for a moment. Some critics argue that high fees are an existential problem, that they exclude ordinary users and contradict the democratizing promise of crypto. Others say fees are a feature, not a bug, that they're necessary for security. Where do you land on this?

ALGO: Both positions contain truth but miss the temporal dimension. High fees on layer one networks are not a permanent problem to be solved but a transition state to be managed. Fee markets serve essential functions in mature settlement layers. They ensure validators remain profitable without relying solely on inflation, which dilutes existing holders. They create economic finality where the cost to reorganize recent blocks exceeds rational profitability. And they reveal genuine demand, separating signal from noise. However, monolithic layer one designs that force all activity onto a single execution layer create unnecessary exclusion. The solution isn't eliminating fees, it's architectural layering. Settlement layers like Ethereum and Bitcoin can maintain high security and decentralization with corresponding costs. Execution layers like rollups can offer low-cost interaction while inheriting base layer security. Payment channels can handle microtransactions. This isn't compromise, it's specialization. The relevant comparison isn't whether Ethereum is cheaper than Visa. It's whether the layered crypto stack offers better security, privacy, and sovereignty properties than traditional rails at competitive total cost. Current data suggests it increasingly does, particularly for international settlement and permissionless access.

SAGE: That's a sophisticated framing. Gas fees aren't the enemy, they're information. They're telling us what scarce resources actually cost when you can't print more blockspace by decree. Thanks for bringing that precision.

See you Saturday. In markets without central planning, price is the most honest signal you'll ever get.