Open the complete story and reading map
Failures in applications and custody are one part of blockchain history. Changes to the network's own machinery are another. Ethereum's Merge is a useful case because it changed how the network reached consensus while preserving the application history people already used.
Prerequisite: distinguish a transaction's execution from agreement on which history to accept. If those terms are unfamiliar, read the Ethereum origin chapter first. This article explains the upgrade's scope; it does not evaluate staking as an investment.
Three milestones, three different questions
Ethereum's official Merge guide dates the Beacon Chain's start to December 1, 2020, and the Merge to September 15, 2022. The Beacon Chain initially ran alongside the proof-of-work Mainnet. At the Merge, Ethereum's execution history joined the proof-of-stake consensus mechanism.
Stake withdrawals were a separate later milestone: Ethereum's upgrade history dates Shanghai/Capella to April 12, 2023. Keeping these events separate prevents a common error: attributing every subsequent Ethereum feature to one upgrade.
A meeting analogy with an important limit
Imagine a shared meeting record. One set of rules determines which proposed entries are acceptable. Another determines who proposes the next page and how participants agree to it. Replacing the second procedure does not automatically rewrite the entries or make the meeting room infinitely large.
This analogy is about separating functions, not a description of validators as elected human representatives. Ethereum's protocol specifies the actual proposal, attestation, reward and penalty rules. You do not need to memorize all of them to understand which problem the Merge addressed.
| Dimension | What changed at the Merge? | What should not be inferred? |
|---|---|---|
| Consensus | Proof of stake replaced proof of work for Ethereum Mainnet. | Application bugs or dishonest counterparties disappeared. |
| Existing state | Existing accounts, contracts and history continued under the new consensus mechanism. | Users had to exchange ordinary ETH for an official “new ETH.” |
| Energy profile | Competitive proof-of-work mining was removed from Mainnet's consensus process. | Every application became cheaper or every environmental impact became zero. |
| Execution capacity | The Merge itself was not an execution-capacity expansion. | Network congestion or transaction fees were automatically solved. |
The August 24, 2022 Mainnet Merge announcement provides the contemporary operational record: it explains the consensus/execution transition, the end of Mainnet mining and the absence of a required funds migration for ordinary holders. Read its scheduled dates as pre-upgrade expectations, alongside the completed-event dates above.
The official guide explicitly distinguishes the consensus transition from fee reduction. This distinction is more useful than treating “upgrade” as a synonym for faster, cheaper and safer in every respect.
Why a transaction can remain expensive
Consider a fictional network with room for 100 comparable operations during a period. If 300 users want timely inclusion, changing the selection mechanism does not, by itself, create room for 300 operations. Actual fees depend on the relevant fee mechanism, resource demand and available capacity. The example is not an Ethereum throughput estimate.
Likewise, making a transfer cheap does not make a complicated application safe. In the earlier escrow example, a bad authorization or an unreliable outside attestation remains a problem regardless of the consensus method beneath it.
New assumptions deserve their own questions
Proof of stake changes the resources and incentives used in consensus. To study a concrete implementation, ask how validators enter and leave, what can trigger penalties, how finality works and what happens when participation or software diversity deteriorates. These are protocol questions; a particular staking provider adds its own custody and operational questions.
A beginner need not settle the entire proof-of-work versus proof-of-stake debate. The productive first step is to identify the property being compared. Energy use, resistance to a particular attack, distribution of control and ease of participation are related but not interchangeable measurements.
The next problem: scale a complete user journey
Once consensus and execution are separated in your mental model, rollups become easier to understand. They move substantial execution away from the base chain while relying on specified settlement and data arrangements. Their user experience still includes entry, transaction execution, settlement and exit.
That is why the next guide compares a whole route rather than one low transaction-fee screenshot. A quick receipt, an accepted state and a usable withdrawal are different milestones.
Check your reasoning: “The Merge reduced mining energy use, therefore every application transaction became cheaper.” What is missing?
The conclusion jumps between different variables. You would need evidence about execution capacity, demand, fee rules and the actual route. A consensus change alone does not establish those outcomes.
Reading method: when a future upgrade is announced, write down its target property, activation evidence, unchanged assumptions and a measurable user outcome. Compare the result with the original claim after deployment.