Blockchains are not startups
tl;dr: Much of the current thinking about blockchain governance misapplies the paradigms of Web 2.0 to decentralized networks. Some allegedly decentralized governance processes (e.g. forking) actually re-introduce subtle forms of centralization (e.g. VC-backed layer 2 projects). Blockchains are better understood as institutional technology expanding the possibilities of economic coordination on the web. Institutions strike a delicate balance between stability and change. Carefully designed formal governance aims to improve stability, accelerate innovation, and facilitate course correction, all without sacrificing decentralization. Once path dependence takes hold, governance and culture become aspects of a protocol which can’t be easily copied.
I. “Attract and extract”
Recent scandals surrounding Web 2.0 giants have accelerated the migration of brilliant developers into the blockchain space. Traditionally, internet entrepreneurs raise venture funds, work to attract users with shiny digital products, and then apply the leverage of network effects to extract value (either that, or they exit to a large company that will). The vision of a decentralized Web 3.0 offers a compelling alternative to this attract-and-extract business model without sacrificing the glamour of startup culture and shiny user interfaces. What’s not to like about “platform-grade” decentralization?
Although it’s exciting that many are rethinking multi-sided platforms and the future of organizations, the platform-focused lens borrows too heavily from the mental models of previous eras. It limits the scope of decentralization to absolving the sins of Web 2.0 giants, imagining that technological advancement and developer community growth will be the key factors that determine a project’s long-term success. Formal governance, we are told, is better left to nation-states, and we shouldn’t prioritize formal rules which might limit the grandness of a founding team’s vision.
We’ve seen this approach before. The “move fast and break things” model of companies like Facebook and Uber offers a liberating allure for those who yearn to build the future. Winner-take-all markets encourage companies to accumulate tens or hundreds of millions of users before addressing the thorny, longer-term issues¹ that require more cautious governance. Even Google doesn’t say “don’t be evil” anymore.
The imperative of decentralized governance
The rise of startup ethos stands in direct contrast to the anarchic state of affairs that has emerged from Satoshi’s disappearance.² Absent a formal mechanism, Bitcoin is evolving into a store of value and so increasingly resembles digital gold. Innovation is shifting instead to VC-backed startups building on layer 2. Perhaps one person’s anarchy is another person’s power vacuum. But anarchic governance does drive positive network effects by giving Bitcoin the latitude to mean anything to anyone and transcend social constraints.
The solution to anarchy, according to many projects in this space, is to apply the classic logic of centralized startups. This, it is alleged, will ensure that protocols stay at the technological frontier. Blockchain protocols do after all bear obvious similarities to startups, and particularly to firms governing multi-sided platforms.³ Like firms, protocols now compete for developers, users, miners (or validators), and capital. Teams led by charismatic founders and brilliant engineers corral network participants with clever go-to-market strategies, partnerships, and operational execution. Private firms building new protocols are often incentivized with founder rewards or early token wealth, encouraging them to improve the protocol, at least within their finite vesting schedule.
However, the traditional startup playbook offers the wrong frame of mind to those seeking to build permissionless open source protocols free of third party control. Too few observers notice the long-term centralization and path dependence that are re-introduced by founder-centric governance. Core protocol developers often claim that their ambitious long-term roadmaps do not sacrifice decentralization, because dissidents can simply hard-fork to create a competing version of the protocol. But this argument ignores that today’s informal governance models give founders an outsized influence over the legitimacy of potential forks, especially since only one fork can keep the original protocol name.
Thus, the original protocol’s name could someday find itself attached to software which has strayed far from the original decentralized vision for which it is known. The protocol may come to benefit concentrated interests at the expense of users. Ultimately, blockchain communities want to stay together to preserve network effects, but absent a formal coordination mechanism, only those who can credibly threaten to hard fork have a meaningful say in deciding a protocol’s future direction. As a result, the fork-centric model often encourages opportunism by centralized third parties at the expense of decentralization (more on this later).
Some characteristics of firms are indeed desirable to reproduce via blockchain governance. Coordination is inherently difficult in a decentralized network lacking the hierarchy of firms to reduce transaction costs and increase predictability. Strong formal assurances (e.g., constitutionalism) and social norms on immutability and inflation improve a protocol’s likelihood of coordinating developers, users, and capital onto the platform without sacrificing decentralization.
We have come to think of computing and storage infrastructure as cheap and widely available, but decentralization forces us to reconsider these as scarce resources (at least for now). In this respect, blockchain governance has many commonalities with the governance of an economy — and just as in traditional economic governance, properly allocating capital to sustain innovation is paramount. But we should refuse to sacrifice decentralization to accomplish these goals.
II. Blockchains as institutional technology
As Nobel Prize-winning economic historian Douglass North outlines in his seminal 1991 piece, institutions are “humanly devised constraints that structure political, economic, and social interaction.” Institutions take the form of both informal constraints (sanctions, taboos, customs, and traditions) and formal rules (constitutions, laws, property rights).
Throughout most of human history, institutional limitations, particularly reliance on informal constraints, bounded human activity. Economic activity began as local exchange within dense networks of hunter-gatherers where trust was high and social enforcement of agreements was untaxing (North). Eventually, a complex international economy with high levels of specialization and productivity arose, necessitating formal rules and hierarchies enforced by nation-states. This allowed parties which didn’t trust one another to initiate transactions over long distances. The rise of formal rules also constrained the arbitrary power of rulers, creating strong assurances.
On the internet, the difficulty of transacting with strangers has reemerged. So have arbitrarily powerful rulers — i.e, web 2.0 platforms and censorship. Blockchains like Bitcoin address these problems by reducing and even removing the need for third-party enforcement of property rules and governance on the internet.
To use North’s terminology, blockchains are internet-native institutions which expand “the opportunity set” of economic activity on the internet. As institutional technology, blockchains may allow us to expand the complexity and scope of value exchange, govern scarce resources, and resolve disputes natively on the internet. Blockchains offer a design space for building and experimenting with institutions native to the internet, but until now there have been few ways to govern these decentralized institutions from within the protocol itself.
Path dependence: a balance between stability and change
North goes on to describe institutions as balancing stability and change. Formal rules are one of the primary means by which humans create stability. However, the adoption of rules may also cause certain entities — namely, those that owe their very existence to the rules — to obstruct even the most obviously beneficial changes. For example, when the government creates intellectual property rights, the decision-making process may come to elevate the interests of patent holders. These patent holders then hinder efforts to experiment with rules abolishing certain patents in favor of prizes. It’s costly to create and adjust to new rules and so entities both explicitly and implicitly work to prevent changes. For the institution, this reflects a phenomenon known as path dependence, in which previous decisions or events limit future decisions. Eventually, it becomes nearly impossible to adopt the optimal set of rules.
The concept of path dependence arose from scholars’ attempts to explain why economies get locked into inferior technological paths and why sub-par technologies can gain network effects over superior ones (looking at you, QWERTY). Given the outsized role of random historical events and the resilience of informal institutions, properly accounting for path dependence is especially of interest in building a sustainable blockchain governance process. We should seek to encourage a stable environment for users, developers, and capital, while minimizing the costs we impose on our future selves in the form of instability or forgone innovation. This applies not only to the protocol itself, but also to the rules governing the upgrade process. It becomes harder to change rules once users form expectations about current rules.
Some issues (e.g., block size) are relatively minor compared to the instability they’ve generated in the blockchain community (e.g. Segwit 2x). In hindsight, many of these issues may seem quaint, but the journey of the blockchain space to date does somewhat resemble a roller coaster ride. Is it just fate?
Nearly all instability originating from within the blockchain space stems from the lack of an obvious ex ante governance mechanism within decentralized networks. Most protocols have chosen centralized governance processes where only entities with credible threats of forking have clear influence. Institutional commitments in this model are thus difficult. As Daron Acemoglu has written, “those in power, e.g., the rulers, cannot commit to not using this power–as long as they do not relinquish it.”
Ultimately, contentious forks represent the greatest threat to the stability of blockchains, and they’re especially problematic for protocols encouraging experimentation on layer 2. The Ethereum Foundation’s Alex Van de Sande made this point in discussing the downstream impact of contentious forks (emphasis added):
Ethereum is more than tokens of course, it’s also a platform for apps, games and other sorts of experiments, and all of them would also be split in a similar way. If you are playing a board game, then you’d suddenly be playing two identical games concurrently, and would have to keep playing both or forfeit the winnings in one of the chain. If you own rare online cats, now every one of them will have an evil twin in a parallel universe.
Likewise, protocol communities will need to decide on topics such as inflation and immutability, as the strong assurance against seizure remains a core part of the blockchain value proposition. Current protocols remain far from being able to offer the level of stability that would be needed to entice most of the world into caring about decentralization. But finding ways to increase stability without introducing centralization and path dependence are paramount.
Change (and how you pay for it)
The blockchain protocol’s own underlying software is a form of digital commons, for which few individual users have sufficient incentive to unilaterally finance improvement. The situation more closely resembles a collective action problem than the classic startup attract-and-extract playbook. Nearly all users of decentralized protocols would benefit from improved protocol functionality, but unlike employees of a firm or members of a small tribe, users face massive coordination costs in the absence of a formal on-chain mechanism.
Proposed funding models for protocol improvement vary. Although token endowments (usually founder rewards) are prevalent in newer projects, emerging models include treasuries funded by block rewards (e.g. Dash), community funds, and inflation funding (minting new tokens to pay for upgrades).
How a protocol pays for upgrades influences decision-making incentives. Like a nation-state managing its public finances, treasury systems funded prior to the decision-making process may encourage projects to misallocate or overspend on behalf of narrow interests. Inflation funding, by contrast, aims to strike a sustainable balance between cautiousness and innovation. To work optimally, the model likely needs to enforce high standards (e.g. formal proofs, futarchy) by which to assess a proposal‘s likelihood of adding value. Unlike a tax, dilution is not felt directly, and so users will likely allocate generously to proposals which can credibly improve the protocol.
Theoretically as the protocol grows in value, even small dilutions (e.g., 0.1% of a large number) will attract competitive proposals among talented developer teams, improving the protocol at low cost relative to the network size. This, as designed, may create a self-reinforcing cycle of decentralized improvement.
The mechanism for protocol upgrades also deserves scrutiny. Reliance on infrequent hard forks encourages core developers to bundle updates into each fork, whereas a formal governance mechanism would enable more incremental upgrades. More importantly, reliance on hard forks implicitly favors the vision and long-term roadmap of the founding team, as only the forks they back are likely to be deemed legitimate.
All forms of governance produce path dependence, but formal governance mechanisms in which users hold final decision-making rights more strongly facilitate a long-term change of course. In founder-dominated projects, reversing path dependence could require departure of the founding team itself. Additionally, independent developers pursuing foundation grants and community funds tend to coalesce around the core team’s long-term vision, reflecting their own reliance on community gatekeepers. As a tacit, evolving process, this is hard to evaluate cleanly today, but it may be introducing hidden path dependence which users will regret later.
III. Blockchain governance in the wild
The past year has highlighted the unmet need for a protocol which can both improve the stability of participating in the ecosystem and facilitate smooth upgrades (i.e. change) without sacrificing the precious decentralization which brought us all here in the first place.
In Bitcoin’s case, the lack of formal mechanism has left the contradictory incentives of miners, developers and users unresolved. Even changes to simple parameters produce rancorous discourse. In this arrangement, entities desiring stasis are incentivized to pursue conflict with those desiring change, further increasing their influence when dissidents hard-fork to focus on their own protocol. Developers who leave to create their own core protocol then often reintroduce centralized governance on the new fork, often naming the fork based on a feature Bitcoin Core has struggled to natively adopt. Ultimately, only the wealthiest Bitcoin users have proper incentive to finance the future of the protocol. This is pushing most innovation to layer 2, where development is more permissionless. The most exciting layer 2 efforts are undertaken by development companies funded by many of the same VCs who financed Web 2.0.
Using layer 2 projects to decentralize funding of protocol improvement is particularly salient in the case of Ethereum. Most of us now associate ERC20 tokens with 2017’s tsunami of token offerings, but back in 2014 Vitalik Buterin spoke excitedly about how “sub-tokens” (ERC20s) would unleash a powerful incentive model for the funding of public goods⁴ on the network (and beyond).
The project’s emphasis on making smart contracts as easy to write as common web applications has dramatically widened the range of developers who can enter the space, prompting a wave of projects modeled after Web 2.0 startups. Ethereum has to some extent decentralized the funding of protocol improvement by enabling nearly anyone to create an ERC20 token, but this has produced real social costs, namely alarming misallocations of capital and over-utilization of Ethereum’s common resources.
Ethereum’s core protocol governance, although public in many respects, remains centralized and not formalized within the protocol. However, the community does increasingly appear to follow norms and precedents (creating Schelling points⁵) established at critical junctures (e.g., the DAO hack). Reinforcing these informal constraints off-chain are many of the ERC20-based layer 2 projects enabled by Ethereum’s approach. Teams of ETH-incentivized developers (ERC20s are usually funded with ETH) have a greater role in protocol governance than miners and VC-backed startups do in Bitcoin governance.
Ethereum founder Vitalik Buterin expressly rebukes on-chain governance, instead proposing “multifactorial consensus.” In this model, core developers continue to control governance decisions, but decisions are informed by an array of layer 2 and off-chain coordination signals representing community preferences. Among these signals are coin-voting and Sibyl-resistant user polling. On some level, all of this aligns with Ethereum’s broader commitment to layer 2 innovation, and some in the community are working to implement these layer 2 signaling tools.
Ethereum researcher Vlad Zamfir’s writings and comments have emphasized that Ethereum’s governance model is perhaps best conceived of as an informally evolving process, and that it is thus much harder to categorize or grok than proposed formal governance schemes. However, some of Zamfir’s observations also reveal the protocol’s path dependence. Even if many in the community wanted to, so much legitimacy is now tied to the technical road map and existing governance processes that it would be hard to change course.
Tezos is a fresh blockchain designed to facilitate secure smart contracts and evolve through formal governance. Nodes will decide upon protocol upgrades through a delegative (liquid) democracy model, directly implementing the chosen protocol amendments. In contrast to existing blockchains, Tezos’ design aims to reserve forks to emergencies. To incentivize innovation, Tezos uses inflation funding to reward upgrade proposals within the protocol explicitly.
Many in the blockchain space misconstrue Tezos as merely aiming to reintroduce the inefficient decision-making processes of nation-states. Many are quick to compare voting to the centralized, technocratic governance of other protocols and perceive voting to be the distinctive feature of formal governance. Rather, Tezos’ use of formal rules must be considered in the context of its overall design. The voting mechanism is merely one part of a broader system incorporating functional programming, Proof of Stake, and inflation funding. As Kathleen Breitman has argued, there probably just isn’t a pre-existing organizational model which by itself is analogous to Tezos.
After all, governance is a much deeper challenge than selecting who makes the decisions: the Tezos approach aims to foster an evidence-based environment that constrains decisions to good and proven opportunities. In the absence of a centralized coordinating entity to filter upgrades by quality, Tezos aims to reinforce good decision-making with formal proofs. These proofs are made easier by the design of its smart contract language, Michelson. Over time, this may also encourage meritocracy, diversify community culture, and reduce reliance on standard gatekeepers. Their code being public, those proposing upgrades need not even be public figures.
Tezos’ emphasis on security and formal governance reflects the core purpose of institutions: stability. By facilitating secure smart contracts and minimizing the risk of contentious splits, Tezos aims to preserve both security and network effects. Instability in these areas significantly hinders capital deployment on current public blockchains.
Yet Tezos’ design also speaks to the imperatives of institutional change. Not only must a blockchain keep pace with innovation by changing its underlying protocol, but it also needs to improve the rules and processes by which those changes are made. As outlined in the Tezos white paper, blockchains are well-positioned to facilitate prediction markets, and thus to enable experimentation with futarchy. This may further improve collective decision-making without introducing a centralized third party.
This is not to say Tezos “solves” governance and Part 2 will dive deeper into the challenges facing on-chain governance.
IV. The lasting power of informality
“What is it about informal constraints,” wonders Douglass North at the end of Institutions, “that gives them such a pervasive influence upon the long-run character of economies?” Efficient institutions rarely arise by conscious choice, but rather evolve unexpectedly from certain informal norms. We often don’t realize it, but informal constraints and social ties also push us to accept governance by inefficient institutions. One doesn’t have to look far. Technology giants may domicile in cloudy Ireland for the low taxes (i.e., attractive formal rules), but they hire where the talent goes, in sunny California. Cities seeking to clone Silicon Valley fail due to a poverty of social capital.⁶
In the context of blockchains, we see this same dynamic in the form of almighty network effects. By this we can explain Bitcoin’s astonishing resiliency as a social construct in spite of technological stasis. Yet we are rapidly hurtling towards protocol interoperability, not only through atomic swaps and synthetic tokens but also through cross-platform smart contract languages. The global, decentralized nature of cryptocurrencies makes it likely that capital and developers will face little to no friction of movement across platforms.⁷ And it is likely that base layer protocols will differ widely with respect to governance model and culture. Once path dependence and historical serendipity take hold, governance and culture will be among the only aspects of a blockchain protocol that can’t be copied.
In the pursuit of truly decentralized governance, experimentation with formal rules remains a worthy endeavor. Tezos reduces reliance on centralized core developer teams, aiming to make the most of blockchains’ potential as decentralized institutions. Restraining the arbitrary behavior of rulers was key to the emergence of international commerce and so too it may be necessary for institutions which can scale to the entire web. We need not sacrifice the dynamism of community and the hope that it inspires. We should hope to decentralize it.
In Part 2, I will address key challenges faced by projects experimenting with on-chain governance.
 One can also think of this in terms of both firm-internal issues, like gender discrimination;  Note that Satoshi’s own disappearance happened as Bitcoin began to grow faster than he liked;  Bitcoin can be modeled as a 5 or 6-sided platform;  Perhaps the real public good provided by Ethereum was easier experimentation;  Precedents are a powerful Schelling point;  Efforts by less culturally dynamic states to lower taxes and attract businesses often function similarly;  It’s worth noting that in proof-of-stake, capital is the protocol.
The Myth of the Irrational Token Holder by Kathleen Breitman
Avoid Evil Twins, by Alex Van de Sande
Institutions and Institutions, Institutional Change and Economic Performance by Douglass North
Why Not a Political Coase Theorem? Social Conflict, Commitment and Politics, by Daron Acemoglu
The Strategy of Conflict, by Thomas C. Schelling
The Long Game in Crypto: Why Decentralization Matters, by Spencer Bogart