Why Ethereum Needs Scaling
Every full node on Ethereum mainnet re-executes every single transaction that ever happens on the network. That design choice is what gives Ethereum its security — anyone can verify the entire chain's history for themselves, with no need to trust a central party — but it also puts a hard ceiling on throughput. Today mainnet processes roughly 15 transactions per second. Compare that to a payments network like Visa, which routinely clears thousands of transactions per second, and the gap becomes obvious: as more people and applications want to use Ethereum at once, demand for a fixed amount of "block space" rises, and so does the price of getting a transaction included.
That price shows up as gas fees. Gas is the unit that measures computational work on Ethereum, and users bid gas prices against each other to get miners (pre-Merge) or validators (post-Merge) to include their transaction in the next block. When demand is low, gas is cheap. When demand spikes — a hot NFT mint, a viral memecoin launch, a major DeFi liquidation cascade — everyone is bidding for the same scarce block space, and fees spike accordingly. During 2021's NFT boom and again during 2023's memecoin rallies, average Ethereum gas fees repeatedly spiked into the $50–$200 range for a single swap or mint. At that price, an application built around micropayments, in-game item trades, or high-frequency social interactions simply doesn't work: the fee dwarfs the value being transacted, and ordinary users are priced out entirely.
A natural question is: why not just make blocks bigger, so more transactions fit and fees come down? This is where the scalability trilemma (covered in depth in the next lesson) first bites. If you raise the gas limit per block, every node must do proportionally more computational and storage work to stay in sync with the chain. Home validators running on consumer hardware — a meaningful chunk of Ethereum's validator set — would get squeezed out, and the validator set would consolidate around well-funded data centers. That is a direct trade of decentralization for throughput, and it is exactly the trade Ethereum's core developers have tried to avoid by pursuing a "rollup-centric roadmap" instead: keep mainnet's base layer lean and verifiable by ordinary hardware, and push transaction execution out to Layer 2s that inherit mainnet's security without inheriting its throughput ceiling. That roadmap is the reason this entire course exists.
Example
Consider a single token swap on a decentralized exchange during a congestion spike. Its total gas cost breaks into three components: the base execution cost of running the swap logic (fixed by the EVM opcodes it touches), the calldata cost of the transaction's input data (historically priced per non-zero and zero byte), and the market-clearing base fee per unit of gas, which is set by EIP-1559 to rise when blocks are full and fall when they are not. During the 2021 NFT boom, a popular mint event could push the base fee from a baseline of ~20 gwei to over 500 gwei within minutes, as thousands of wallets simultaneously submitted mint transactions competing for the same 30 million gas of block space. A transaction that would normally cost $2 in execution fees suddenly cost $60–$100, purely because of contention for a fixed resource — not because the underlying computation changed at all.
total_fee = gas_used * (base_fee + priority_fee)
# gas_used ≈ fixed by the operation (e.g., ~150,000 for a swap)
# base_fee ≈ rises/falls with block fullness (EIP-1559)
# priority_fee ≈ tip bid by the user to jump the queueThis formula shows precisely why "just raise the gas limit" looks like a free fix on paper but isn't: raising the limit increases the gas_used ceiling per block, which temporarily eases base-fee pressure, but at the cost of every validator needing to process and store a bigger chain, long-term.
Exercise
Using a public gas tracker (such as Etherscan's Gas Tracker or a similar historical gas-fee dashboard), pull one week of mainnet base-fee data and identify the three highest-congestion events in that window. For each one, note what application activity (an NFT mint, a token launch, a liquidation cascade, etc.) most likely caused the spike, based on the timing and what you can find reported about that period. Then write two sentences explaining why simply raising the block gas limit is not a free scaling fix — be specific about what resource becomes scarcer for node operators as a result.
Recap
- Ethereum's full re-execution model caps mainnet throughput at ~15 TPS, and fees rise sharply whenever demand for block space exceeds that ceiling.
- A transaction's fee is driven by execution gas, calldata cost, and a dynamically adjusting base fee — not by the "value" being moved.
- Raising the gas limit scales throughput only by shifting cost onto every validator's hardware, which is the centralization trade-off Layer 2s are built to avoid.