Hard Forks on the Bitcoin Blockchain: Reversible Exit, Continuing Voice

Abstract

Hirschman’s Exit/Voice conception has stimulated an extensive literature across many disciplines. Exit/Voice was designed to shine explanatory light on the responses of stakeholders in firms and states in moments of decline. We apply it here to the Bitcoin blockchain, where the Exit/Voice apparatus illuminates strange and unencountered qualities of the reactive choices open to the blockchain’s varied stakeholders.

Introduction

The fairly recent experience of forks and potential forks in response to the various Segregated Witness and blocksize increase proposals highlights the nature of emergent governance on the Bitcoin Network.1 The result of the SegWit/blocksize controversies was a so-called hard fork of the Bitcoin blockchain,2 producing two varieties of the Bitcoin blockchain with a common history but now distinct technical characteristics: Bitcoin and Bitcoin Cash.3 Forks are blockchain-specific events that evoke elements of both  “exit” and “voice” in the sense of the Exit-or-Voice reaction paradigm first set out by Alfred O. Hirschman.4 Hirschman’s Exit/Voice conception has stimulated an extensive literature across many disciplines. Exit/Voice was designed to shine explanatory light on the responses of stakeholders in firms and states in moments of decline. We apply it here to the Bitcoin blockchain, where the Exit/Voice apparatus illuminates strange and unencountered qualities of the reactive choices open to the blockchain’s varied stakeholders.

The Bitcoin blockchain marks a novel form of social organization. The architects of the Bitcoin blockchain claim that it embodies decentralization.5 There is no central node in the network, no center of authority directing or coordinating internal or external action. Rather, the constituent autonomous nodes operate the Bitcoin blockchain following a downloaded open-source protocol that Bitcoin’s mysterious founders initially developed and which is quite resistant to change.6 When change does come to the Bitcoin blockchain, it emerges from loose and informal constellations of various stakeholders. Bitcoin users and the sponsors of the Bitcoin network nodes (known as “miners”) are formal stakeholders. Indirect stakeholders include Bitcoin developers and businesses that service the Bitcoin ecosystem (systems operators and equipment manufacturers, as well as Bitcoin exchanges). Still we can draw a black box around the Bitcoin blockchain and examine it as a finite social space, an organization set apart from surrounding players and institutions.

There is no clear “fundamental” stakeholder in the Bitcoin blockchain. Orthodox legal understandings of the business organization give primacy to the interests of shareholders. Similarly, democratic accounts of political organizations see the people (or demos) as the locus of paramount interests. The Bitcoin blockchain presents a more ambiguous case. The Bitcoin blockchain serves both users and miners; each stakeholder class is essential. But there is no settled view as to which of these stakeholders should defer in the event of conflicting interests.

The Bitcoin blockchain resorts to “Nakamoto consensus” as its ultimate form of decision-making. Nakamoto consensus is an emergent and diffuse accord arising among the active Bitcoin miners, each pursuing its own advantage while collectively engaged in maintaining, verifying and expanding the blockchain. Nakamoto consensus is the source, and hence authority, as to the canonical state of the Bitcoin blockchain; this consensus constitutes the pragmatic Truth7 as far as the Bitcoin blockchain is concerned, and once reached, it cannot readily be disturbed. A consensus forms with a 10-minute frequency as to the validity of a newly promoted block and its insertion into the blockchain. More considered consensus engages on those occasions when the Bitcoin blockchain community makes a constitutional decision as to changes to its basic rules.8

Miners are essential stakeholders in the Bitcoin blockchain; without their efforts, the system would experience network delays or would require abandonment of the proof of work algorithm, which is thought to be essential for the blockchain’s security. The miners toil with their racks of specialized computers and earn Bitcoin in return. The effective compensation for their efforts, which yield Bitcoin over time, depends on earning Bitcoin’s block rewards and transaction fees, the value of which are functions of Bitcoin’s prevailing market prices. Only if Bitcoin is valued will the work of the miner pencil out.

Users, the holders of Bitcoin who transact on the blockchain, are another essential constituency. Users do not “vote” or otherwise engage in the formation of Nakamoto consensus, yet their aggregate demand for Bitcoin drives its market price. A Bitcoin has no intrinsic value, but it does have attributes (scarcity and transferability) that attract users. Bitcoin and the Bitcoin blockchain are one seamless invention: one does not exist without the other, and from a functional perspective, it is not meaningful to speak of a point where one stops and the other starts. Demand by users fixes the Bitcoin market price, which indirectly determines the amount of mining power which will be supplied to the network.

Thus, two groups of stakeholders, miners and users, are essential to the health of the Bitcoin blockchain ecosystem. There are in fact two forms of consensus maintained: one formed by the miners about the “state” and “rules” of the blockchain, and another established by users setting the market value of Bitcoin. A feedback loop9 links these two forms of consensus. Users derive utility from—and support the value of—Bitcoin, which in turn incentivizes the miners. The miners construct the secure, decentralized and tamper-proof blockchain (through the exercise of Nakamoto consensus), earning the users’ confidence.

For purposes of this Essay, we will explore the corporate governance branch of the Exit/Voice discourse and treat the Bitcoin blockchain as a novel form of business organization, where stakeholders participate in the pursuit of wealth maximization. Bitcoin could also be viewed as a political organization, but that would be a different essay. Indeed, one can fall deeply into the Bitcoin “rabbit hole” and once there become so absorbed in the virtual world of Bitcoin and its politics that it might seem as real as the material world of earth, air and fire that envelops the rest of us.

1. Exit and Voice on the Bitcoin Blockchain

1.1. Change Comes to the Bitcoin Blockchain. A third set of stakeholders design and improve the functioning of the Bitcoin ecosystem. Blockchain developers, by prestige and persuasion, guide the evolution of Bitcoin infrastructure and contribute the hard work of developing implementable solutions. Only developers actively code. Developers do not formally touch the blockchain,10 but their influence on its rules and functioning is profound. Anyone who stumbles upon the thick Twitter, Reddit and Medium chatter addressing Bitcoin reform or reads the Bitcoin-related GitHub postings can readily monitor activity in the Bitcoin developer community. Developers (some more than others) wield impressive power in implementing their ideas, notwithstanding the open-source, decentralized ethos that surrounds Bitcoin. These constitutional amendments are called Bitcoin Improvement Proposals (“BIPs”) and are intensely studied, tested and debated.

The developer community can only propose change. Implementation of BIPs requires the reaching of consensus by the other players.11 Formally, miners implement Bitcoin Improvement Proposals through Nakamoto consensus. Each miner weighs the consequences of the decision to adopt a change, with an eye on divining what the broader community is thinking. It is generally in a miner’s best interest to join the camp it projects will command an enduring consensus, even when the miner does not share the belief that the anticipated consensus determination—to implement a BIP or not—is the best future course for the miner or the best for the overall blockchain. A miner’s “vote” is better understood as its prediction of the collective stance of the broader miner community than as an expression of the voting miner’s individual preference.12

BIP 9 was introduced to serve as a signaling mechanism, so miners could indicate that they are prepared to implement a change. For example, some BIPs are designed to “activate” upon the signaling of a super-majority of miners. Yet, miners’ signaling via BIP 9 may reveal supplier preferences which are at odds with consumer preferences. Ultimately, miners qua suppliers yield to demand. Whether a particular change sticks or not depends, in the end, on its effect on the price of Bitcoin, a price that is determined by the aggregate of Bitcoin users, present and prospective.

Changes to the Bitcoin blockchain protocol often proceed smoothly. Developers present BIPs, which are then debated and refined. Once a BIP commands the support of the technocracy, it is put before the miners. Miners signal their support for proposals by raising “flags,” that is, by including expressions of support in an available field found in the header of every block. These flags can be seen and registered upon the revelation of each new block; as individual miners sporadically succeed in sponsoring a new block, they use these opportunities to raise their own “flag” (or not) on a particular proposal. Over a period of time, as more and more miners raise flags of support (or fail to do so), a group position emerges. The number of flags at a miner’s disposal depends upon the mining resources controlled by the mining pool; a powerful mining operation will secure more frequent opportunities to form new blocks, and so have more opportunities to raise flags. Mining pool operators may vote for their clients by proxy. The process is a form of rolling election, although no miner raising a flag is bound to follow through with its declared intention when time comes to give effect to the decision.

Miners truly “vote” by choosing where to build new blocks onto the blockchain. They can build on either an implementing or non-implementing block (this is how forks come about at decision points in blockchain history). Until consensus forms (through the building out of new chain growth, indicating the relative command of hashing power applied to each alternative), flagging is simply predictive—like a voter’s response to a pre-election poll. Further, the observation of a flagging by a broad (or at least representative) sample of miners has no binding effect. Still, strong signals of support for a particular reform permits spontaneous coordination among the otherwise independent miners.

The press and others tally these flags to measure the prospects for a consensus. At a certain point—predestined informally—sufficient signals have been registered to enable the otherwise uncoordinated system to shift smoothly to the new rule. The raising of flags over a period of time, as various miners promote new blocks onto the blockchain, functions as a kind of pre-election polling, gauging sentiment and promoting the formation of a common understanding of what will come to pass.13

However, things do not always work out so smoothly. Some proposals generate division within the community. When consensus fails to form as to the adoption of new rules, either temporarily or persistently, we describe the event as a “fork.” Implementation of some changes will technologically divide the miner community; these are called “hard forks.” A miner who fails to implement the new rules will be effectively excluded (due to technological incompatibility) from continued participation on the blockchain. Non-implementing miners may follow prior rules and form a rival branch of the blockchain. Other changes are gentler; these are soft forks. Soft forks are backward-compatible, in that they accommodate participation by miners who implement the change, as well as those who do not.14

A hard fork may result in two blockchains, each branch commanding the efforts of a sufficient amount of hashing power (provided by adhering miners) to be viable. Each branch of the fork shares the same pre-fork history, but post-fork, each branch progresses on its own way with no regard to the unfolding history of the other. The vast majority of forks are transitory—one breach is abandoned and the entire community re-aligns around the surviving branch, which then forms the canonical blockchain.

Network effects incentivize an eventual convergence of two branches to one surviving blockchain. The currently persisting divide—and co-existence—of the Bitcoin and Bitcoin Cash blockchains is unexpected.15 The continuing co-existence of these two branches may signal the presence of valuable product differentiation; it may also reflect continuing uncertainty as to which set of technological characteristics is the better path forward for the Bitcoin ecosystem.

1.2 Discovering Voice

1.2.1 Discovering Voice: Miners. The Exit/Voice choice becomes relevant when a social organization is in decline.16 Decline can be measured against objective metrics (fixed or relative); it can also be measured against one’s past hopes and expectations. A prospective change can be viewed by a stakeholder as an instance of decline, however measured, and may stimulate an urge to exit. Where Exit is available at no or low cost, it is the dominant response. Exit is only meaningful where there is an alternative position that satisfies the departing stakeholder’s needs. In many solutions, Exit will impose painful costs (economic losses), that induces that “wonderful concentration of mind”17 of others.

And so it is with the Bitcoin blockchain. The SegWit/blocksize episode involved two technologically separate and politically intertwined elements.18 The first was SegWit itself, that would expand room on each block by moving signature information from the body of the block into the block header (it would also solve the quite serious problem of “transaction malleability”). The second was an initiative to increase block size, from its current 1MB limit to something twice to eight times larger. Both of these proposals would address capacity limitations of the Bitcoin blockchain.

Hirschman observes that in many social organizations, stakeholders enjoy the continuing opportunity to participate in the governance of their organization through exercising their voices. Voice includes but is not limited to exercise of formal voting rights.19 In the SegWit/blocksize controversies, miners, users and developers discovered effective ways of giving Voice to their concerns with, or hopes for, the various SegWit and blocksize increase proposals under consideration.

Miners may exercise Voice. In the run-up to the resolution to the SegWit/blocksize controversies, miners could raise flags to signal their respective stances. Or not. When the SegWit amendment was first opened for expressions of support, miners were slow in raising their flags of support.20 Their quiet was understood to signal a reluctance to adopt the SegWit proposal and accompanying block size increase. Silence, it turns out, may be just another form of Voice.

1.2.2 Discovering Voice: Users. Users can exercise pre-decision Voice by the usual means: blogs, tweets, podcasts. They can also exercise Voice by holding Bitcoin, that is, by not resorting to Exit. Indeed, the Bitcoin community has spawned the deliberate practice of “hodling” (both an ironic misspelling of holding and an acronym) Bitcoin, which expresses a commitment to remain engaged in Bitcoin despite the temptations of selling out (and hence heading to the Exit) at an available high price. Users also have an ambivalent role. While they undertake a more passive role as holders and users of Bitcoin (similar to shareholder passivity), they obtain their position exclusively through active secondary market transactions. You can buy your way into holding Bitcoin, but you cannot earn them (unless you have been a successful miner). Users can be said to provide capital to the Bitcoin project only indirectly, as they stand ready to purchase the haul of Bitcoin produced by miners. Miners are necessarily paid in bitcoin, yet their electricity bills are paid in fiat currency. All else equal, the markets must continuously absorb the supply of new crypto-currency to keep the price afloat. Their cumulative presence as a market creates the financial conditions necessary to incentivize the miners to mine.

Users can and do exercise Voice. They write, they tweet, they podcast. The Bitcoin community is a thriving debating society. The views of individual users about the competing SegWit and blocksize proposals were widely communicated. These include scathing appraisals of particular miners and their motives.

Users also arrogated a competence to initiate a fork to settle the SegWit matter: the so-called User Activated Soft Fork or ‘“UASF.”21 In a UASF, certain nodes committed to act to represent the views of a constituency of users and other businesses in the Bitcoin economy in pushing a soft fork implementation of SegWit—where SegWit-compliant and non-compliant blocks would both be processed. The UASF committed to a pressure campaign (that involved a degree of self-sacrifice) that was thought to be capable of tipping the blockchain into full implementation.22 In the end, the UASF was not directly successful in resolving the SegWit debates, because its effect was mooted by acquiescence of the miners.23 Going forward, the promotion of a UASF might be viewed as an effective form of user Voice. The success or failure of a UASF depends upon psychological factors, which will be determined in future soft forks.24

1.2.3 Discovering Voice: Developers. The developer community is formally the least powerful of the three major classes of Bitcoin blockchain stakeholders. Yet it has the clearest access to Voice. The developers after all propose any changes to the blockchain architecture; other stakeholders may wish for reform, but without technical know-how they are unable to give coded form to their desires. Only developers can design and communicate proposed changes. As such, the developers speak and are listened to. Every BIP is authored by a developer. Developers can and do disagree about approaches. But the free-for-all of proposal formation provides developers with a remarkable degree of freedom. Voice is not costless, of course. A developer may need to invest time and money to flesh out an idea so that it can bring the broader community toward its view. Beyond these limits, however, a developer has relatively uninhibited Voice. The presence of Voice is predicted to suppress any urge to exit. No doubt there are developers that have moved from the Bitcoin community to other projects (Ethereum is largely built by former Bitcoiners), but the fact of Exit does not prove the absence of Voice.

With each stakeholder class there will likely be conflicting expression, as individual stakeholder interests are variegated. Large holders of Bitcoin Unspent Transaction Outputs (“UTXOs,” or Bitcoin holdings), known as “whales,” will have different perspectives from recent retail purchasers of Bitcoin. Dominant mining groups—particularly those associated with proprietary equipment—will express preferences different from those of smaller, independent outfits.

Voice exists for each constituency—at least during the pre-decision phase. There are, however, decisions of a constitutional nature that may drive a particular stakeholder to feel any further exercise of Voice is futile. At that point, Exit comes into play.

1.3 Reversible Exit.  Stakeholders also possess, to a varying practical degree, the ability to exit an organization, by either selling their stake (such as shareholders selling shares) or abandoning it (such as citizens emigrating from the territory or a polity). Similarly, in the SegWit/blocksize episode, principal stakeholders expressed, through a range of subtle responses, their willingness to exit. Users may at any moment sell their Bitcoin; the blockchain is open to them 24 hours a day and merely requires their signatures to effect liquidating transactions of their UTXOs. Likewise, given the open nature of the Bitcoin blockchain network, miners are free to go. They need not sign off; miners can simply detach and apply their hashing power to other cryptocurrency projects with the same mining algorithm or may pause mining operations (and save on electricity costs). To the extent miners hold stocks of Bitcoin earned through prior activity, their ability to maintain these holdings is unaffected by any determination to exit the Bitcoin blockchain qua miners. And developers can port their experience and expertise to other projects.

Once users sell, they cannot readily return to Bitcoin. Regret will likely be dear. Unless a selling user had been fortunate (having sold “high” and later able to buy “low”), reversibility of Exit will be costly. Despite acute downturns on occasion, the overall upward price trajectory of Bitcoin to date has been dramatically steep; if it continues, a departed Bitcoin user may not be financially able to recapture his prior position. For users, an Exit from Bitcoin may be definitive. We may reach a day when the Bitcoin price stabilizes; that may create a greater chance for a regretful user to return to Bitcoin.

Miners have a much easier time returning. For the moment, both the Bitcoin and Bitcoin Cash branches of the original Bitcoin blockchain are open network. A regretful miner can simply reappear on the scene, download the current version of the relevant Bitcoin blockchain and resume activity. For miners, Exit is reversible; indeed, miners have a continuing free option (as we all do as prospective miners) to join the open Bitcoin networks without regard to any past history.

1.4 Continuity of Interest.  It now seems obvious with hindsight, although it had not been clear earlier to many users, what happens to Bitcoin in the event of a hard fork. The not-so-intuitive result has been that users can now locate their pre-fork Bitcoin (UTXOs) on each of the two resulting blockchains. This seems, perhaps, like a windfall, but financially a fork is experienced more like a corporate restructuring, where an established shareholder receives shares in each of the now-independent firms resulting from a split. In theory, the sum of the shares of the surviving entities should approximately equal the value of the pre-split firm (unless one believes in the Wall Street magic that creates value out of mere changes in form). The same financial logic should apply to cryptocurrency resulting from a fork; the sum of the value of each surviving coin (Bitcoin and Bitcoin Cash) should equal the value of the pre-fork coin (Bitcoin).

2. Exit in the Complex Social Space of the Bitcoin Blockchain

2.1. Exit for Miners. The social space defined by the Bitcoin ecosystem is populated with stakeholders that do not match the better-explored divide between management and capital. Miners in fact devote capital to the blockchain through their purchase and deployment of specialized equipment and their payment of electricity costs. Miners host the blockchain. Their rewards are the Bitcoin they earn in promoting new blocks to the blockchain. Like independent contractors, miners are hired by the network to provide security.

Miners exercise voice by engaging in Nakamoto consensus. At the technological level, miners select which block to build on. At a hard fork, a miner faces a choice between two competing blocks, one on each side of the fork. The miner’s choice as to where it will devote its continuing energies constitutes the exercise of its “vote”; in some sense, this choice is the most powerful form of Voice known on the blockchain. Collectively, the choices of individual miners as to where they focus their efforts determine which branch of the fork will thrive.

Miner Exit is more nuanced than the departure of a shareholder. A miner need not find a purchaser to exchange positions; a miner exits by simply ceasing to mine. There is no forfeiture of previously earned Bitcoin; these may be held or sold as the miner sees fit. A departing miner’s Bitcoin sell/hold policy can remain exactly as it had been without regard to Exit. Miners may come to own sizable amounts of Bitcoin, but at least for the moment, the possession of a stock of Bitcoin—or lack—does not carry any weight in establishing their status as miners. Total deployed hashing power, and not stakes of Bitcoin, makes a miner a force in the community.

Miners do come and go. The Bitcoin blockchain is an open system. Participation as a miner requires no prior authorization; a new miner simply downloads a complete copy of the blockchain and the requisite software and goes to work. When miners depart, they simply “hang up,” disappearing from the network without prior notice or explanation. Importantly, a miner can leave the Bitcoin blockchain, mine elsewhere for while (there are other mining opportunities in the cryptocurrency space) and then later return. For the miner at least, Exit is reversible.

A hard fork presents a peculiar form of miner choice, which displays aspects of both Voice and Exit. As a miner selects which fork to pursue, he communicates a choice. The value proposition for mining one cryptocurrency for another (the choice a miners faces upon a hard fork) depends on at least two factors: (1) the total amount of hash power devoted to the particular divide under consideration and (2) the value of the coin received by a successful miner.

The amount of migrating hash power results from an uncoordinated and aggregated decision. At the moment of its decision, a miner will not and cannot know how many other miners will end up following one branch of the fork or the other. As post-fork mining progresses, the relative speed of puzzle solution, given the difficulty of the puzzle target, will reveal the cumulative amount of hash power that followed the fork.

The value of the respective coin is revealed as well as time progresses. Bitcoin Cash may be easier to mine, but its value is a fraction of the value of Bitcoin. While the miner faces uncertainty immediately after a hard fork, when an initial decision must be made, the miner can fairly quickly determine whether the mining proposition remains profitable, at least in the short term.

Miners who “exit” Bitcoin to pursue mining on Bitcoin Cash can at any moment flip back to Bitcoin.25 Exit, for miners at least, is reversible at low cost. The post-fork history seems to demonstrate that some miners simply mined the more profitable blockchain. Other miners remained on the same network for political reasons. Bitcoin Cash’s “emergency difficulty adjustment” (“EDA”) produced large gyrations in the hash rate of the network. More recently, Bitcoin Cash has implemented a mining algorithm which readjusts the mining difficulty on a six-block rolling basis. Bitcoin continues as before to readjust its difficulty level every 2,016 blocks.26

Since Bitcoin Cash’s difficulty readjusts on a continuous basis, its blocks will average ten minutes regardless of its overall hash power. Thus, there is less of a marginal benefit to adding additional hash power to its network. Indeed, additional hash power would further insulate the network from a 51% attack, which can currently be performed by a minority of Bitcoin’s mining power. Due to Bitcoin’s longer-term difficulty readjustments, there is a greater marginal benefit to adding hash power. In the short run, the supply of hash power is inelastic. In the long run, the supply of hash power is elastic.

2.2. Exit for Users. Users can also exit the Bitcoin ecosystem in a familiar way: they can sell their Bitcoin. Indeed, given the once-stratospheric heights of Bitcoin prices, Exit might have seemed like a prudent choice regardless of the SegWit outcome. The eventual SegWit/2x hard fork that divided Bitcoin from Bitcoin Cash effectively doubled (in a nominal sense) the user’s holdings. A particular quantity of pre-fork Bitcoin gave rise, by operation of this fork, to a holding of that same quantity of post-fork Bitcoin and that same quantity of Bitcoin Cash.27

Without the need for active election, Bitcoin users followed both forks simultaneously. The post-fork Bitcoin and post-fork Bitcoin Cash both trace to the same “Unspent Transaction Outputs” recorded deep within the common pre-fork history of both blockchains. But those same historic UTXOs can be spent independently on each blockchain. After a hard fork, users must engage in at least two sale transactions to completely exit the combined Bitcoin space. Or they can exit one blockchain and retain UTXOs on the other. Hence the 2 x 2 matrix of post-fork sell/hold possibilities: hold one, sell the other; hold both; sell both. Curiously, it seems that the combined value of the post-fork Bitcoin and Bitcoin Cash did exceed the value of the pre-fork Bitcoin. This suggests, but does not necessarily demonstrate, that the SegWit fork created value; that the market prefers the availability of choice between these two flavors of Bitcoin. Bitcoin and Bitcoin Cash markets then as now are quite thin, so one should not read too much into any instantaneous price.28

2.3 Exit for Developers and Other Stakeholders. Developers can freely depart the Bitcoin blockchain, and many have.29 There are other engaging projects in the blockchain and cryptocurrency space where developer talent is highly prized and easily placed. Developers can, of course, also return; in this, they (like miners) can exercise a reversible exit. Of course, the exchange of harsh words and pride may make a return to the Bitcoin fold unlikely.

Manufacturers of mining equipment “exit” in the sense that they may drop a specialized product line if they perceive that a particular evolutionary path is doomed.  This is essentially a commercial decision. The Exit of a supplier is the symmetric twin to the Exit of the customer discussed by Hirschman.

3. Schism

3.1 Schism and Re-Convergence. A hard fork, whether viewed from the perspective of a miner or of a user, is both Voice and Exit, and neither. It is something of a hybrid not reflected in Hirschman’s description. Imagine that the relevant stakeholder views either an improvement proposal or the status quo to constitute decline. A hard fork opens up both the desired and the dreaded pathways. So long as one of the fork branches promises an acceptable outcome to the stakeholder, he can follow that branch and discard (that is, exit) the other branch.

Again, the experience of a hard fork differs between miners and users. When facing a hard fork, a miner must act. Persistent hard forks result when the miners are divided; they arise where (1) some miners prefer the new rule and others prefer the status quo, and (2) both groups are, at least in the short run, each able to devote sufficient processing power to sustainably support a separate branch of the forked blockchain. Given the reality of two branches, a miner must allocate his hashing power between one and the other.

There is a technological aspect to this choice. A miner with a variety of mining rigs might devote certain rig types to one branch and other types to the other, depending on the match between machine capabilities and the contrasting hash power demands of the two branches.

The miner who makes the all-in choice with respect to one of the two branches will participate in the formation of a Nakamoto consensus with respect to that branch. This is Voice. Indeed, it is a maximization of Voice, as it is concentrated on that single branch. And that same miner’s abandonment of the other branch, leaving it with diminished hashing power, is also a form of Voice.

Moreover, the all-in miner has exited the abandoned branch. But recall this Exit may be reversed so long as that branch continues as an open network. The miner cannot avoid distributing its pre-fork hashing power between the two branches, exiting one branch or the other (or part from one branch and part from another) with its pre-fork hashing power. Every mining rig has to be directed—at any given moment—to only one blockchain; it cannot serve two masters.30

A hard fork to a miner is like a religious schism. A community is fractured by a pair of mutually incompatible rules that require an adherent to take a stand, in or out, explicitly with one and implicitly with the other.

The user faces a quite different choice. The user experiencing a hard fork will have its coins (UTXO) recognized on both surviving blockchains. Both coins have market value (though their value can and do differ significantly). The user need not exit either solution. She can continue to hold both forms of coin.31 Exit by the user is not forced, but it is available. And again, the user can sell either coin, or both. Most holders of UTXO continue to hold both Bitcoin and Bitcoin Cash. This reflects unawareness of the hard fork, existence of lost private keys, the effects of inertia, or a hedging strategy.

Developers are specialist, and they undertake discrete projects. One imagines that developers will have a purer experience of schism. A developer is not only looking at which branch is likely to survive; he may find one branch more attractive from a technical standpoint. Moreover, he may locate his development efforts on the branch that is more receptive to his ideas. As the code-bases diverge, fewer innovations will be susceptible to “copy and paste” between projects.

3.2 Scale Effects in the Face of Schism. Schisms are destabilizing and potentially value-destructive. Scale effects operate to discourage forks. In the case of a hard fork within the Bitcoin community, each branch requires a certain degree of infrastructure independent of the other branch. Forks create inefficient redundancy of infrastructure. They also divide the developer community into smaller camps—which may or may not retard the rate of innovation—as well as rewards to be derived from innovation. And as two branches share, at least in the short run, a quantity of nodes that had earlier been devoted to a single blockchain, there is likely some diminishment of security.

The SegWit/2x fork increased the nominal supply of Bitcoin-like coins, from a maximum of 21 million Bitcoin prior to the fork to 21 million of post-fork Bitcoin and 21 million of Bitcoin Cash. Of course, the value of all coins should not have significantly increased—unless the now differentiated Bitcoin and Bitcoin Cash networks, by presenting the market with differentiated products, unlocks value. Even so, the combined value of 21 million Bitcoin and 21 million Bitcoin Cash must outweigh the structural redundancies introduced by the fork in order for their co-existence to be economically stable.

Scale effects should favor a single winner from the Bitcoin/Bitcoin Cash rivalry. The current relative value of the coins may reflect the bookmaker’s odds that one or the other be that survivor. The eventual survivor should capture the current total combined market value of these two, highly similar coins. Were Bitcoin Cash to suspend operation today, miners would migrate to the Bitcoin blockchain. Users of stocks of pre-fork Bitcoins would see the value of their corresponding Bitcoin Cash holdings evaporate and the value of their post-fork Bitcoin appreciate to a compensating degree, with little or no financial loss. Only post-fork acquirers of Bitcoin Cash would experience losses; those users would not likely be able to outrun the accelerating loss of value triggered by a dawning general recognition that Bitcoin would be the sole survivor of the two branches. And if Bitcoin changes its proof of work algorithm, then many of the SHA-256 ASICs32 will migrate to the Bitcoin Cash chain. This has given Bitcoin miners a “back-up plan” in the event they are ousted from the Bitcoin network.

4. Loyalty and Continuity of Interest

In Hirschman’s conception, Exit generally dominates Voice. Where Exit is relatively costless, a stakeholder will depart an organization that it views to be in decline. Voice rises in importance where Exit is occluded. Exit is a fairly open course to users, miners and developers, so the presence and energetic exercise of Voice is, at first blush, anomalous.

Hirschman describes the strange reaction of remaining engaged in the presence of the possibility of Exit and gives various accounts for it. One who has the chance for Exit in the presence of decline yet chooses to stay is displaying Loyalty. Pursuing a loyal stance is risky, as the loyal stakeholder cannot know with confidence whether her continuing presence—an exercise of Voice—will bring about a positive improvement.

Stranger still are displays of Loyalty by stakeholders who have departed an organization and who remain engaged in influencing its destiny. These departed stakeholders exercise Voice from outside. Hirschman argues that such loyal departed have not in fact executed a complete Exit; their continuing engagement signals interest (perhaps in the nature of an interest in the continuing production of a public good by the organization).

A user who rides both branches of a hard fork does not display divided loyalties. Post-fork users may wish to retain holdings of UTXOs on both branches as a matter of diversification, spreading risk between the two positions. Or the user may perceive distinctive utility from each position. A circulating story is that Bitcoin functions better as a store of value, while Bitcoin Cash is better suited for small payments. A user could desire both access to both functionalities. A user that exits its position on one branch or the other, by selling coins, may continue to exercise Voice, by participating in the overall conversation or as a potential market re-entrant. Loyalty to the larger Bitcoin project may engage users beyond their private interests.

Miners may also display Loyalty and exercise Voice notwithstanding a decision to terminate mining on one branch or the other. Miners face both near costless Exit and near costless re-entry: the reversible Exit described above. Reversible Exit creates an option and an option will usually have some positive value. Given this option value, miners may appear to act loyally to one or both branches of the soft fork by remaining engaged and exercising Voice.

Conclusion

The Bitcoin blockchain provides an intriguing setting to apply and adjust the Hirschman Exit/Voice paradigm. It features a complex of stakeholders that do not correspond, in function or interests, with the familiar players found in corporate governance contests such as shareholders, directors and management.

Blockchain stakeholders experience governance in different ways. Major governance events frequently take the form of hard forks. A successful hard fork will produce a continuing chain that incorporates the new rule or protocol, but a second branch of that chain may also survive (at least for a while). The fairly recent hard fork that produced the split between Bitcoin and Bitcoin Cash is an example of such an event.

The blockchain-specific hard fork gives users a range of Exit responses. Users may sell Bitcoin or Bitcoin Cash or both. Moreover, a user can continue to “hodl” coins on both branches of the post-fork blockchain. And regardless of the hodl/sell determination, users can and do continuously exercise Voice as to policies on both branches. A selling user may subsequently purchase a position and re-enter; the selling users’ reserve price for re-entry potentially contributes to the market price of both Bitcoin and Bitcoin Cash. A user who believes that only one branch will survive in the longer run, but is not clear which, will be motivated to exercise voice in promoting rules and policies that will preserve and grow the Bitcoin ecosystem.

Miners face the freest course: they can exit and return from one branch or the other at little cost or inconvenience. This possibility of reversing Exit may promote a more continuing interest in the exercise of Voice. Miners, at the moment, seem particularly interested in supporting policies that will promote the maximization of transaction fees. These interests will be expressed with regard to proposals for both branches, irrespective of which branch a miner may be actively working at the moment.

____________________

The Authors are grateful for the research assistance of Ruth Kwon.