Ethereum & Layer-2

One dimension measured against many

A simulation compares a post-Glamsterdam single base fee with EIP-7999 configurations, and finds the baseline permits 288–403 GiB of state growth a year.

2026-09-21·3 min·Source: Ethereum Research

One lane, and no way to say you can wait

Transactions differ in how their value decays with time. Ethereum prices all of them with a single scalar fee, and reads the result as a capacity problem.

2026-08-28·3 min·Source: Ethereum Research

A committee of 512 votes on the payload

Mike Neuder walks through ePBS as it stands in EIP-7732: the proposer signs a bid and the builder publishes the payload three seconds later.

2026-08-24·3 min·Source: Ethereum Research

A timing game that costs honest validators 15%

A proposer who broadcasts late makes honest attesters vote for the wrong head. A distance-weighted reward is proposed to cut the loss to 4.71%.

2026-08-20·2 min·Source: Ethereum Research

A data availability proof checked by a machine

leanDA's cryptographic core now has a Lean proof of completeness and binding, covering the optimisation that made it practical.

2026-08-18·2 min·Source: Ethereum Research

Glamsterdam gets a testnet you can join

Platåberget runs for months rather than days, and it breaks any tool that assumes a fixed maximum gas limit.

2026-08-17·2 min·Source: Ethereum Foundation

Four ways to price more than one resource

EIP-7999 asks how execution, state growth and data can carry separate base fees behind a gas interface that only knows one number.

2026-08-12·3 min·Source: Ethereum Research

Rollups stopped competing on cost and started competing on distribution

When fees are a fraction of a cent everywhere, the differentiator is who already has the users.

$0.003▼ −41%median L2 transaction cost

2026-08-10·10 min

Sequencer revenue is the business model, and it is centralised

The part that earns is the part that has not been decentralised.

$96M▲ +12%annualised sequencer revenue, top five

2026-08-06·6 min

Client diversity improved; operator concentration did not

Two different concentration risks, frequently conflated in the same chart.

38%▼ −3ppstake on largest client

2026-08-02·7 min

Blob pricing found a floor and stopped being interesting

Supply exceeded demand for most of the year, which is the point.

0.9▲ +0.2average blobs per block

2026-07-29·5 min