Guide
Why the client mix matters
What a supermajority client bug costs at 33% and at 66%, why the published diversity figures disagree with each other by more than they agree, and what any of it changes for an application rather than a validator.
What a supermajority bug costs
Client diversity is usually presented as a civic virtue. It is really a description of what happens when a majority client has a bug, and the consequences escalate at two specific thresholds.
A bug in a consensus client with over 33% of the Ethereum nodes could prevent the consensus layer from finalizing, meaning users could not trust that transactions would not be reverted or changed at some point.
a critical bug in a client with a two-thirds majority could cause the chain to incorrectly split and finalize, leading to a large set of validators getting stuck on an invalid chain.
Those quotes are about Ethereum, and the mechanism is not. Any chain with one dominant implementation has the same exposure — which is why BNB Smart Chain narrowing to two supported clients and Solana gaining a second implementation are the same story told twice.
What it changes for an application
For a validator, a supermajority bug is a slashing risk. For an application it is subtler and closer to home: it is the scenario in which your provider, its competitor and your own node are all wrong in the same direction at the same time, and every fallback you built agrees with the others.
- Diversity is what makes independent failure possible, and independent failure is the only thing redundancy can exploit. Two providers running the same client on the same chain are one provider with two invoices.
- You usually cannot find out which client a provider runs, which is most of why we started asking every endpoint and publishing the answers.
- The choice you do control is your own node. If you run one, running the minority client costs you very little and is the only contribution to this that an application operator can actually make.
Then read the number’s method before quoting it
Client share is measured, not known, and the measurements disagree — by more than they agree. These three are published side by side on the same page.
| Source | Lighthouse | Prysm | Teku |
|---|---|---|---|
| Blockprint | 0.16% | — | 99.83% |
| Miga Labs | 50.86% | 20.97% | 7.97% |
| Rated | 20.6% | 21.17% | 53.86% |
Which is worth stating plainly on a site made of measurements: this is the same discipline every figure here is held to. A number carries how it was collected, over what window, and from how many samples, or it is not publishable — the methodology exists so that our own figures can be interrogated the way these ones just were.