A Philosophy of Subtraction
How the Ethereum Foundation drives innovation through delegation
In September, I promised that one of our goals for the Reboot newsletter would be to uncover “how the sausage gets made”—teasing apart the marriages and tensions between theory and practice, peeking behind the scenes at how technologists understand the ethics and impact of their work.
This incredibly researched essay by Saffron Huang is one of the best (and only?) analyses of the Ethereum Foundation, one of the most quietly influential yet important organizations shaping crypto today. It spotlights their professed “philosophy of subtraction,” an unusual approach for a frontier technology organization, and one worth juxtaposing with the VC-driven growth gospel that dominates industry today.
P.S. Today Saffron launched The Collective Intelligence Project’s first whitepaper, detailing an R&D agenda for a “public goods-oriented model for technology [governance and] development.” Check it out!
A Philosophy of Subtraction
The ubiquity of growth
The most striking thing about the Ethereum Foundation, the Swiss non-profit that was created to support the Ethereum blockchain ecosystem, is that they hold “subtraction” as one of their three basic principles. (The others are “long term thinking” and “stewardship of values.”) In a world where organizations intentionally or unintentionally tend towards greater scope, the EF’s approach is, distinctively, one of self-minimization.
The EF was created to enable and coordinate the development of Ethereum, a decentralised and programmable public blockchain. The organization received 8.3% of all Ether (the token that powers the Ethereum network) created in the initial crowdsale. Aided by the greatly-appreciating value of Ether, EF has financed itself with this treasury ever since—doing things like making software, giving grants, and maintaining educational resources.
Given the technology it supports, the EF’s approach of subtraction feels fitting; just as Ethereum is meant to be decentralising, so does the organisation behind it also seek to “[resist] the natural tendency of organizations to grow and accumulate value within themselves,” instead attempting to cultivate value in the broader ecosystem. They claim to do this by seeking to distribute opportunities, rather than capture them; to be happy rather than territorial when other similar organisations create value they could have generated; to matter less, not more.
In an age that prizes exponential returns, maybe this “philosophy of subtraction” sounds ridiculous. Our canonical metrics never yearn for shrinkage. A country’s doing well if its GDP is increasing; a company is thriving if it increases shareholder returns and captures greater market share. Everyone seems to want to go from 0 to 1, or from 1 to 100—to bring something into existence, and thereafter make it as large as possible. We believe that bigger is better, whether it’s burgers, impact or Net Promoter Scores. The tech industry is particularly obsessed with winner-take-all growth. Who better than Peter Thiel, the instigator of the 0 to 1 meme, to insist in his book that every successful business is a monopoly?
Even non-profits, ostensibly without the same financial imperatives to expand, often depend on grants from wealthy foundations and other benefactors, pushing them to market the size of their impact, take on additional work to meet stipulations, and hire more staff.1 Rather than solving a problem and thereby working themselves out of a job, such incentives make non-profits dependent on securing larger grant streams in the future, bowing to the same growth logic. Predictably, the foundations to which many non-profits look for funding also tend towards accumulation. The 200 largest US foundations on average earned 7.62% return while paying out 4.97% annually between 1972-1996 – spending essentially the minimum legal percentage and actually increasing their endowments overall.2
Maybe it’s something like a loose law of human nature: once formed, organizations generally like to grow: to perpetuate their existence as an end in itself, accumulate resources, and become ever-more important to—and depended upon—by the ecosystem.
Of course, there are logical reasons for any single actor to focus on growth. Economies of scale is one practical reason. But bigness can be costly. Institutions get bloated, too big to fail until they do. And at a large scale, many things are hard to do well: Facebook finds it nigh impossible to properly moderate the activity of billions of users. The ecosystem often suffers too—very large corporations are hard to contest, and end up holding concentrated and unchecked power.
Recognizing the downsides of a growth culture, many people are looking for alternative economic models. In addition to developing new macrostructures, we should also pay attention to organisations operating on counter-growth philosophies in the here and now, contributing to a more pluralistic economic system. While none can be a panacea, and new efforts generally have rough edges, we can still learn from bold, imperfect attempts at prefiguring something different.
What does the philosophy of subtraction look like in practice?
Josh Stark, who leads operations at the EF (he says he’s like the COO, but they don’t have titles) makes sense of the philosophy of subtraction in four ways. The first is that it resists a natural process of organizational accumulation, and excessive identity- and ego-building. Humans like to feel needed; we want to defend the work of ourselves and our teams, and we want to build up skills viewed as important and in-demand. “Subtraction” recognises that this impulse is quite normal, but needs to be kept in check.
The second lens is that a philosophy of subtraction is a process of pushing things outside oneself; a vision of an institution that distributes the money, expertise, power and prestige it has collected to others. The third lens decenters the organization as an agent – rather than assuming they’re the primary actor when confronted with a problem, the EF tries to ask, “How can the ecosystem fix this, and how can we support that? Could some other organisation(s) be more fitting for the task at hand?”
The final lens is that subtraction is not (just) a theory of action; it is a vision of a future: A future where a more distributed, pluralistic set of smaller organizations coordinate to serve vital functions, in Ethereum and elsewhere. This outlook reaches beyond the competitive-but-consolidated status quo in many markets, where oligopolies dominate industries including telecommunications, social media, and food manufacturing. Subtraction is a vision that prioritizes coordination more than competition.
Formally, the EF organizes itself in a four-layered structure to carry out this philosophy, where they give themselves less power in each successive layer of their organisation. These layers are: in-house teams, ecosystem grants, delegated domain allocators, and third party funding. The further out the layer, the more that funding and decision-making is allocated to external parties. You can imagine seed ideas that get incubated at inner layers and pushed outwards when better external homes are found for them, or projects that move inwards if more central coordination becomes required.
The first and most centrally-controlled is “in-house teams”, which does things like building the Solidity programming language and organizing the EF’s official conference. This layer is essentially what most organizations do — carry out tasks within internal teams. Stepping one layer out, “ecosystem grants” means is how the EF delegates money to outside teams. For example, they gave grants to five external teams to build “consensus clients” necessary for the latest Ethereum upgrades. The next layer is “delegated domain allocators”: teams that work with the EF on higher-level decisions for how to give funding. For example, Devfolio, an organisation that grows the software community in India, has helped the EF make decisions about how to grow the Indian Ethereum ecosystem. Lastly, the most “subtracted” layer is “third party funding.” As the ecosystem has matured, the EF has started giving more money directly to external groups like Nomic Foundation, ETHGlobal and 0xPARC, to make grant allocation decisions without any EF input.
Interestingly, the EF also cares about how much soft power they capture. There used to only be one website for both Ethereum and the EF, a comprehensive source for learning about both the protocol and the organisation. But according to Aya Miyaguchi (the EF’s Executive Director), they decided to split out ethereum.org as a separate open-source resource and guide to Ethereum that doesn’t visibly belong to the EF. By contrast, the EF’s online presence at ethereum.foundation is minimal.
Josh also explains that, when making internal decisions, the EF considers who else is doing what in the ecosystem so that they avoid becoming too dominant in the space. That doesn’t mean the EF does nothing—that’s being subtractive in a superficial way—instead, he believes in contending with “narrow and rather granular” trade-offs. The EF has carried out this philosophy even in high-stakes settings, where declining control risks long-lasting consequences. As mentioned, the EF asked five different teams—Prysm, Lighthouse, Teku, Nimbus and Lodestar, names faintly reminiscent of spaceships—to build consensus clients rather than building one in-house. The consensus clients are fundamental infrastructure for Ethereum’s recent upgrade to Proof-of-Stake, so there’s a solid argument that it would have been better to build one internally, in close collaboration with other core EF engineers to ensure they didn’t introduce bugs into a chain that secures $200 billion of assets. However, they decided to subtract, giving themselves a coordination role instead of directly building the software. This decision also gives users more choice of which client to use, increases the overall system’s resilience through redundancy (what if a client has a bug in it?), encourages competition and opens the door for external stakeholders to make decisions on core upgrades. As a caveat, this outsourcing would not have been possible when the Ethereum ecosystem was less mature, so the EF’s decision also signifies that the ecosystem is more ready for more trust.
In most industries, including tech, a decision like this is exceedingly strange; imagine if OpenAI asked five external teams to build different ChatGPT interfaces for the same underlying model rather than making one themselves, and let users pick whichever one they wanted. Or if Canon asked five other companies to build the lenses for their camera bodies. Even in most open source software projects, organizations build the components that they need rather than taking the extra effort of coordinating multiple other projects for the sake of their principles and long-term vision.
Is the EF’s philosophy of subtraction working?
Now, is their philosophy of subtraction actually working? Is the EF getting less important?
This is hard to figure out. How would we measure their progress on subtraction? Maybe the organisation can track the same metrics that others do to measure importance, like press articles and social media mentions, but aim to have them go down over time. But good metrics for subtraction don’t seem straightforward; influence and impact are slippery measures to grasp.
For an external observer, part of the difficulty of evaluating this is that the EF stays relatively quiet about themselves—this is meant to be another subtraction-oriented idea, but it makes analysing them difficult.
While the EF is generally respected within the Ethereum community, some folks I talked to had the opinion that they were actually growing in importance, given their ambitious near-term roadmap. A core contributor that works at Consensys (which does R&D and builds Ethereum products and tools) mentioned that there’s been a big upscaling of protocol-level ambition in Ethereum, and that while the recent Merge saw Ethereum’s electricity consumption drop by over 99.9%, it is not the biggest thing on the horizon. They have grand plans to also implement the Surge, Verge, Purge and of course, the Splurge—all forthcoming upgrades for improving the scalability, sustainability and security of the chain. The size of R&D teams at the EF has definitely increased; while they disbursed 20 million in external spending in 2021, they spent 28 million internally.
And really, even if the EF is asking external teams to do the work, they still play a central role in coordinating the different players, not to mention their influence and importance in providing funding and direction for the grantees in their orbit. The ecosystem structure is far more of a hub-and-spokes model right now rather than a distributed web: the spokes are connected to the hub (the EF) rather than to each other.
Josh says that their perspective on subtraction is a relative, not an objective one. The EF has grown, but Ethereum has grown dramatically, so they are trending towards mattering relatively less. And even if it’s true that the EF hasn’t decreased their influence as much as they will say or hope, the long-term ambition of subtraction, and their existing efforts at implementing it, are already rare and noteworthy aims for a frontier technology organization.
One clear downside of subtraction is the speed tradeoff. For example, because the EF supports five different external consensus layer clients, people end up as coordinators of multiple heterogeneous parties instead of acting as managers operating within a single, institutional culture. This slows things down, and increases the opportunity for chaos and misalignment. Some more traditional investors have commented that the EF should be structured more like a normal company, so that rather than “whittling the value [of Ethereum] away,” they can “build this thing into a monster.”
However, subtraction might still be faster in the long term. There’s an implicit promise that while the EF is slow for now, they’re growing a diverse set of overlapping institutions that lead to ecosystem resilience and sustainable coalitions. They won’t make rash, totalizing decisions that will get them taken down by regulators, or fail as a company due to interpersonal conflict, or be as susceptible to downtime. The point of a public blockchain is to be infallible, and the implicit promise of stability from the EF invites people to build faster-moving applications on top of the steady rock of the technology. The technical reliableness holds true: Ethereum’s never been hacked, and as far as anyone can tell, the Merge upgrade happened perfectly.
Would this approach be effective for other organisations?
Organisations that want to subtract one’s presence and externalize decision-making are unusual across all sectors. The stated goal of philanthropy is to give all your money away until you can disappear, but that isn’t what really happens in most cases. It’s getting more popular, though – the ill-fated FTX Future Fund had a regranting program, one example of a fund delegating decision-making power to external individuals. And even the Gates Foundation has a mandate to spend all its money within 20 years of Bill and Melinda’s deaths.
If subtraction were more of a default, there would likely be smaller and more organizations in a given ecosystem, including some that play coordination roles (to decrease the transaction costs of dealing with external entities more frequently). The optimistic outcome is that we’d have more players, more choice, dynamics of co-creation rather than competition, and an overall more resilient and pluralistic ecosystem.
To create and maintain such an ecosystem, we’d need the right norms and sustained motivation. Subtraction seems to require mutual goodwill between players, or an organization (like the EF) that has an abundance of options, cash reserves, and isn’t hampered by scarcity. EF’s core developers and leaders are not, in general, being rallied against, so they can set a subtraction roadmap that promises a controlled and smooth ride. Simply put, other people see them as trustworthy, and this matters—despite the professed trustlessness that their technology enables. It’s probably because Ethereum is not a blatant cash grab by its founders; Vitalik is good at setting a tone of openness, optimism and goodwill. And many EF employees have worked substantially below-market wages for years in order to contribute to the core protocol.
Organisations that hope for widespread adoption of their work, e.g. those that are oriented around setting protocols and standards (e.g. Fido for authentication or Farcaster for social media), may be well-suited to a very similar philosophy of subtraction. Protocols and standards matter insofar as you have a lot of people using them. Being credibly neutral and minimally involved, increases trust and therefore adoption. If you want to grow a sustainable, thriving ecosystem too, you’ll necessarily need to encourage external organizations to come in, make decisions, and capture some value. If Internet protocols like TCP/IP had been invented by a private company, they might have been competed against, or not gained the same amount of widespread adoption.
For tech companies, does the Thielian narrative of aiming for monopoly have to hold, or can you adopt a philosophy of subtraction, too? Can you build highly valuable, potentially resource-intensive technology similar to Ethereum, but still be cooperative and perhaps even philanthropic? It seems less straightforward, but more businesses could integrate aspects of this vision, particularly when seen as creating positive-sum partnerships. Subtraction means more opportunities for collaboration with others with different comparative advantages.
One example is Craigslist, which offered countless services, many of which eventually became competing, often venture-backed and more upscale businesses, like Fiverr, Etsy, iDate, Coursera or Stubhub. Yet Craigslist stuck to its core purpose rather than attempting to capture all the value it could have by becoming an Amazon-like juggernaut. The overall ecosystem of peer-to-peer exchange products is likely more pluralistic, and better off for this choice.
A subtractive tech company could look deliberately to build for underserved gaps in the market rather than areas of overhyped competition, do its core purpose extremely well, celebrate the wins of other companies, and be suspicious of default expansion. It could function as more of a busy bazaar than a top-down-built cathedral; a Schelling point for coordinating activities in the broader ecosystem, and a place where community input is leveraged for better decision-making.
Particularly value-driven leaders could even be happy to dissolve their businesses and when they achieve their mission, or when they decide the world no longer needs them. We often assume that if something’s good and useful, it’ll last forever — but that doesn’t seem to make much sense. Surely, some goals get reached at some point, after which we can all pack up and move on. The team at Convergent Research, which are launching a new non-profit research organisation model called Focused Research Organisations (FROs), specify on their website that FROs are “concerted, finite-duration (5-7 year) efforts, after which the FRO disbands and deploys its technology into the real world.”
A significant underlying problem is to find alternatives to venture capital funding structures. Venture capitalists push portfolio companies to focus more on growth than profit, because the only ways they can return capital to investors are from the company going public or exiting. Companies are pressured to scale as big as possible so that they can go public or catch Google’s eye for a buyout, which encourages growth for its own sake. We need better ways for companies to generate revenue from a thriving ecosystem that they helped to create—e.g. emphasising zebras over unicorns, or innovating on what term sheets or venture funds can look like. The EF has an unusual advantage here; given its treasury, it doesn’t have to navigate the standard tech funding pressures. Levelling the playing field for other organisations, in this regard, is important.
There are clearly not enough organizations oriented towards goals like this in practice, and plenty of work to do in imagining alternative institutional goals and structures. But I have great hopes that we can start testing and building a plurality of viable organizational structures that are better aligned with manifesting the world we want to see.
Saffron Huang is a technologist, writer and researcher. She is co-director of the Collective Intelligence Project working on scalable democratic methods for governing technology, formerly a research engineer at DeepMind, and co-founded / was the creative director for Reboot’s Kernel Magazine v1.
Disclosure: Saffron received a grant from the Ethereum Foundation in August 2022 for research on centralisation in the crypto ecosystem. She did not receive EF funding for this Reboot essay, and has no ongoing relationship with the EF.
Reboot publishes essays and interviews reimagining tech’s future every week. If you liked this and want to keep up, subscribe below ⚡️
The Substackhas said its final goodbye, but it's always worth a flip through their archives—the kind of mini-ethnographies, weird internet paraphernalia, and hard-hitting analysis you won’t find anywhere else.
💝 closing note
From the community:
Nate Lane wrote “HTML Is For All of Us”: on how Web 2.0 gave us user-friendly interfaces in return for central corporate control, and why technical literacy might be the key to a more democratic web.
One more plug to check out the CIP whitepaper if the ideas in this essay excite you. The team is looking for collaborators!
Jasmine & Reboot team
INCITE! The Revolution Will Not Be Funded: Beyond the Non-Profit Industrial Complex. Duke University Press, 2017. Accessed 28 January 2023.
Deep, Akash, and Peter Frumkin. “The Foundation Payout Puzzle.” Taking philanthropy seriously: Beyond noble intentions to responsible giving, 2006. 189.