Lauran

Farewell, Ethereum#

On November 16, 2024, it was the second-to-last day of Devcon, the largest official conference for the Ethereum ecosystem held in Bangkok. The Devcon team stated that this was the most attended Devcon ever, with over 12k participants.

On the same day, Péter Szilágyi, the team leader of Geth, the largest execution client on Ethereum, posted a vacation notice on X.

In the comments of this tweet, Ethereum founder Vitalik, Ethereum Foundation director Aya Miyaguchi, Ethereum protocol team coordinator Tim Beiko, and several EF researchers expressed their gratitude and best wishes.

Vacation is an ambiguous excuse, and the atmosphere in the comments section does not feel like a normal expression of appreciation for a member on a regular break.

Afterward, Péter Szilágyi added a year (2015 - 2025) to his personal profile on X, seemingly conveying the message of his departure.

From the start of his vacation until June 2025, there has been no clear, public conclusion about whether Péter has left. Most of his content on X has been about hardware device explorations, interspersed with retweets and congratulations regarding new personnel appointments at EF since 2025.

On June 3, 2025, EF released an announcement about the reorganization of the research and development team, with several teams being disbanded or restructured.

On June 10, Péter made a rare appearance expressing dissatisfaction with Ethereum after his vacation: he mentioned that around 2018 to 2019, he had applied to EF for unconditional back pay for the authors of key dependencies for Geth, such as a $10k back payment to the author of go-leveldb, but EF could only offer $500 in unconditional back pay, and to receive a larger amount, a contract promising deliverables was required. During the same period, EF allocated $5 million to the new client Parity (OpenEthereum) without any additional conditions. Péter stated that EF's explanation for this action was that Ethereum needed more clients and did not believe Geth would remain. The announcement of the grant can still be found on the EF official blog here.

On the same day, June 10, Péter commented on the recent disbandment of several Ethereum teams, stating that he had spoken with the newly appointed executive director Hsiao-Wei Wang, who said, "Geth will also be removed in the coming years; EF will only focus on research and education."

An account named Ethereum ARM retweeted and questioned: Remove Geth? For Ethereum, we truly hope this never happens. If there is a project worth funding, it is this one.

Another newly appointed executive director, Tomasz K. Stańczak, urgently responded to this post: There are no plans to remove Geth. It is a great client software and a talented team contributing to protocol security. We will maintain/support Geth and continue to make it better and faster.

However, this comment provoked a more intense reaction from Péter, and the story behind his vacation last November began to emerge.

In response to Tomasz's statement, Péter posted multiple times revealing the following information:

1. EF offered Geth $5 million to split off and operate independently.
2. EF asked the Geth team three times if they were willing to split; Péter, Felix (now the Geth team leader), and another member opposed it.
3. EF initiated (and funded) a second Geth team within Nethermind, which according to EF team member Josh Stark, was a "100% independent fork of Geth with no intention of collaboration." In November 2024, less than 24 hours after Péter and Geth team members discovered this "secret Geth" team, Péter was fired, but at the time it was publicly announced as a vacation. The reason for the dismissal was: "Threatening to leave is unacceptable and will undermine team morale."
4. In the weeks leading up to early June, Tomasz K. Stańczak, who held dual roles as EF executive director and Nethermind founder, contacted most of the remaining Geth developers, telling them to start interviewing at other companies because he believed they should not be paid so much, and asked who would leave, cutting their salaries in half.
5. EF had requested Péter to return in February, March, April, and on June 10, but Péter insisted on an apology, which was deemed impossible, so he also refused to return.

Under this series of posts, executive director Hsiao-Wei Wang commented: We are discussing the future; this is the only thing we can do for you.

On June 12, in response to numerous questions, Tomasz opened an AMA post on X.

Péter asked a question in the comments, seemingly provocatively:

Oh, oh, oh, I have one, pick me, pick me: If (to quote you) "a mistake was made," why doesn't the EF leadership take responsibility while the developers are being dismissed? Is there any framework to hold higher-ups accountable, or is this just Vitalik's rule of law?

Tomasz denied the claim that the management team does not take responsibility: I disagree with your view that the EF management (when you say "EF" here, what do you mean - it sounds like you might be referring to the management/specific individuals) does not take responsibility. We have implemented changes to the Protocol team. As a result, some people have left. They may all be excellent and will quickly find new roles in the ecosystem or their expertise. Some of them may have been involved in the wrong projects or temporarily lost motivation, as often happens when certain changes are delayed. I am monitoring the performance of the management team. Hsiao Wei and I are not responsible for the board, but I am responsible for ensuring the delivery of other members of the management team. I am satisfied with many of the changes we have introduced together since March. We are willing to continue receiving more feedback. We are also empowering more team leaders, who will learn and improve rapidly as they take on more responsibilities now.

Péter continued to reply, directly targeting Aya and Josh: I am talking about Aya and Josh; to my knowledge, most EFers would be happy without them. But you certainly have always known this:)

Tomasz replied very officially: Aya is a board member and is not part of the management team. Josh is one of the best operators I have worked with. Both are great people.

Péter bluntly stated that the responsible parties were not the two current executive directors, but someone else: I am not talking about you and HWW; the purpose of your hiring is clearly to take on the losses of the previous batch. I hope they are held accountable, not you.

Tomasz replied: I hear you. I understand you are suggesting I should dismiss the management team. Based on the information and experience I currently have, I disagree with your view.

This round of dialogue regarding accountability in the EF management team came to an end. From the information currently known, there are two nodes in the past decade where Péter has had significant disagreements with EF as a member of the Geth team:

1. 2018 - 2019: EF did not trust Geth's loyalty and disagreed with providing unconditional funding.
2. 2024: EF funded and secretly formed another Geth team and proposed to split Geth.

Regarding the former, based on the comparison of funding received from Geth and Parity, the fact is that EF did not provide Geth with unconditional funding. As for whether EF really funded a new "Geth" within Nethermind, Tomasz responded in an AMA post and attached a proposal link on Magician: It is not a team created to somehow undermine the main Geth repository. This is a small grant for 1.5 people at Nethermind to separate L2-related changes into the Geth codebase.

This proposal mentions that EF and Nethermind jointly created a Geth optimized for L2, filling the gap where Geth does not serve L2. Currently, this proposal is not intended to replace Geth; the Geth mainline continues to focus on L1 execution clients, pursuing security, stability, and broad compatibility, while rollup-geth is aimed at L2, allowing sequencers/provers to iterate quickly as needed.

On June 12, during a Geth AMA event at Protocol Berg in Berlin, Geth members faced some rather pointed questions. When asked about Péter's departure and public criticism, they acknowledged Péter's contributions but stated that some of his "conspiracy theory-like" criticisms of EF (such as "EF wants to secretly replace the Geth team") were not true.

In summary, the secret formation of a Geth team by EF may be an exaggerated description, but it could also be a profound insight.

Geth: A Decade of Growth#

The original vision of Ethereum was to become a world computer, and this vast computer requires countless nodes, each containing at least one execution client and one consensus client. Geth is currently the most widely used execution client. EF funded multiple teams in different cities to develop three independent protocol implementations in the early days: Geth (Go language), Pyethereum (Python language), and C++ implementation. Geth was developed under the leadership of the EF core team, with early key developers including Jeffrey Wilcke. Under the recommendation of an early participant in crypto, Péter Szilágyi came into contact with EF and gradually became the team lead for Geth. Over time, Geth became the dominant client due to its stability and performance.

If a client exceeds 33% market share, it raises centralization concerns. Since its inception, Geth's market share has consistently remained above 30%.

Starting in 2015, EF began to fund projects like Parity, Nethermind, and Besu. The Parity that Péter Szilágyi mentioned receiving $5 million in funding later transitioned to a DAO organization for maintenance when the Parity Technologies team shifted its focus to building the Polkadot ecosystem, ultimately announcing the cessation of the project in May 2022 and collaborating with Erigon to complete user migration.

In 2021, EF launched a client incentive program that provided rewards in ETH to client teams as long as they continued to build software that met mainnet performance and security requirements. The rewards would unlock over time, with a total reward of 4608 ETH.

After a decade, the result of EF's diversification of execution clients is the emergence of five relatively mainstream execution clients: Geth, Nethermind, Besu, Erigon, and Reth, with Geth accounting for 41% of the market share (as of August 2, 2025).

(Data source: https://clientdiversity.org/#distribution)

The four execution clients Nethermind, Besu, Erigon, and Reth are operated by organizations outside of EF, equipped with professional management personnel, and have more diversified funding sources.

In contrast, Geth is highly dependent on EF for both management and funding (see chart below).

If Geth were to split off, it would need to consider how to operate sustainably as an independent infrastructure beyond just technical development and maintenance, which involves considerable market and financial work. The former is something Péter Szilágyi has consistently refused to develop in Geth; he even felt it was unnecessary at the Devcon venue. Financial work requires professionals, meaning team members would shift from purely technical roles to a mix of technical and non-technical roles, increasing management difficulty. Péter candidly stated that his refusal of the split proposal was because: We are simply not good at running a company, and we do not have the corresponding infrastructure or team support; the whole thing will ultimately fail.

Beyond management, what other concerns does Geth have?

Funding. Structurally, EF hopes to focus on research and education in the future, gradually splitting off the implementation parts, giving Geth $5 million, and transforming Geth from a direct subordinate relationship into two independent organizations. But how long can $5 million sustain Geth's operations? According to EF's November 2024 report, L1 R&D accounted for 25.7% of funding in 2023, approximately $34.7 million, allocated to Geth, Solidity research, the Devcon conference, and the Ethereum Robust Incentives Group. On average, each project could receive $6.94 million, which is the minimum amount Geth might spend annually. In reality, given Geth's importance as infrastructure, it might receive a larger share than an even distribution. Therefore, the promised $5 million from EF for the split could likely only support Geth's technical development expenses for one year after independence. What happens after that? If, as Péter stated, EF did not believe Geth would stay in Ethereum and thus was unwilling to provide unconditional back pay, Péter would inevitably doubt that the split-off Geth could secure other grant support from EF. Thus, the $5 million could very well become Geth's last supper.

Internal conflicts of interest within EF. From an organizational perspective, EF's proposal to split Geth while secretly funding the construction of a "Geth" within Nethermind feels like a betrayal to the Geth team. Current executive director Tomasz, who holds dual roles at Nethermind and EF, brings to mind last year's incident where Dankrad Feist and Justin Drake served as EF researchers while also advising EigenLayer. These two situations are not fundamentally different; whether as EF executive directors or EF researchers, both roles require significant conflict avoidance. However, this time, EF's personnel restructuring and funding of new execution clients did not involve conflict avoidance, making the situation quite suspicious. While EF funding other execution clients is acceptable to Péter, as he is neither the founder of Geth nor a maximalist, having advocated for and encouraged diversity among execution clients on multiple occasions, the actions of EF in dismissing Péter under the pretext of "threatening to leave undermines team unity" and Tomasz informing other Geth team members to seek other opportunities, etc., inevitably cast a shadow of suspicion over the matter.

EF is undergoing a transformation, from reshuffling management to adjusting research teams, moving towards an organization focused on education and research. From this strategic goal, splitting Geth aligns with the transformation. However, this conflict is not merely "the growing pains of reform"; it is not necessary.

An infrastructure that has operated on the foundation of Ethereum for a decade is undeniably important. Compared to the need for transformation, assessing Geth's actual role in the ecosystem and how to keep it operating sustainably and stably is the most crucial issue. However, from the management's various actions, there seems to be a unique bureaucratic indifference, accelerating forward without regard for others.

On July 23, Péter posted his speech at ETHCluj on X, stating it was his last talk about Ethereum, discussing the origins of decentralization and the current centralized state of Ethereum. At the end of the tweet, he bid farewell with a line from The Hitchhiker's Guide to the Galaxy: So long, and thanks for all the fish! Péter left Ethereum, and some projects are eager to recruit him; perhaps he will embark on a new journey and continue to actively critique everything on X.

As for Geth, given EF's shift, it seems destined to split. How EF and the current Geth team coordinate to help this decade-old Ethereum organization transition out of its childhood, become independent, yet remain loyal to Ethereum, is a question. After all, the current team has also clearly expressed opposition to independence in the AMA, as Geth cannot be profitable and is essentially serving the Ethereum protocol.