Greg Markou, who maintained web3.js during the London hard fork, says testnets can’t catch every edge case that breaks live apps
Greg Markou has a warning for anyone who assumes Ethereum upgrades go smoothly because they passed a testnet: they often don’t.
The co-founder and CEO of Sprinter says upgrades to the network are hard to test and can break a large number of applications without warning. He speaks from experience, having been in the room the last time it happened.
His comments came during a public discussion among developers about the realities of shipping protocol changes on the world’s busiest smart contract platform. His central point was blunt. Testnets help, but they cannot account for every edge case that might trip up apps running on the live network.
A London hard fork war story
Markou’s credibility on this topic comes from a specific incident. He was the maintainer of web3.js during the EIP-1559 upgrade, which shipped as part of the London hard fork in August 2021.
For the uninitiated, web3.js is a JavaScript library that lets apps communicate with the Ethereum blockchain. Think of it as the translator sitting between a website and the network itself.
A hard fork, meanwhile, is a change to Ethereum’s core rules that the network adopts as a whole. Every node has to follow the new rules, which means any software built on the old assumptions can suddenly find itself out of step.
According to Markou, an update tied to that upgrade had consequences nobody saw coming. Nearly every application that relied on web3.js was affected.
The culprit was an invariant involving MetaMask that went undetected. An invariant is a condition that developers assume will always hold true. When one of those assumptions quietly stops being true, everything built on top of it can wobble.
Markou described Ethereum as an “absolute monolithic beast.”
It is a vivid description, and not an affectionate one. The point is that the system has grown so large and interconnected that predicting how a single change ripples outward is genuinely difficult.
Why testnets aren’t the safety net people assume
Testnets are practice versions of a blockchain where developers can try changes without risking real money. They are the dress rehearsal before opening night.
Markou acknowledged that testnets are beneficial. His argument is narrower: they fail to cover all the edge cases that can hit live applications.
On mainnet, apps interact with wallets, libraries, and other contracts in combinations that no single test environment fully reproduces. The MetaMask invariant that tripped up web3.js users in 2021 is a case study in exactly that kind of blind spot.
Who Markou is, and what Sprinter is building
Markou previously co-founded ChainSafe Systems, the firm that maintained web3.js. Sprinter launched out of ChainSafe, and he now leads it as CEO.
The company raised $5.2 million in a seed round in November 2025. Its focus is cross-chain credit and infrastructure, with a pitch it calls “solving-as-a-service.”
Sprinter also plans a native governance token called SPRINT.
No new upgrades or protocol changes have been introduced as a result of the discussion. This is a conversation about lessons learned, not an announcement.
Disclosure: This article was edited by Vivian Nguyen. For more information on how we create and review content, see our Editorial Policy.