Chapter 05: The Bitcoin custody paradox

Here is the June 2026 draft of chapter 05 in my forthcoming new book, When Policy Falls Behind: Bitcoin, AI, and the Governance of Fast Systems. Comments are welcome! Click here for a downloadable pdf version.

Introduction

Bitcoin's custody paradox arises because of the nature of agency in Bitcoin’s permissionless environment. A pension fund allocator that acquires Bitcoin exposure through a US spot ETF holds a regulated financial claim on Bitcoin rather than a protocol position – the actual UTXOs – in Bitcoin itself. The keys to access Bitcoin sit with another institution and the allocator cannot choose the client software implementation its investors endorse, maintain the infrastructure through which implementation level preferences become credible, or reposition during a governance dispute without fiduciary review. The fund has acquired exposure while surrendering the form of agency that self-custody makes possible.

The tradeoff becomes more consequential as Bitcoin exposure migrates from direct bearer holding into regulated financial products. Legacy financial governance makes that migration appear natural because large pools of capital require auditable custody, fiduciary process, regulatory recognition, reporting infrastructure, and identifiable intermediaries. Those arrangements lower transaction costs and make Bitcoin legible within portfolios that could not plausibly manage private keys directly. Financialization therefore solves a real access problem, but at the cost of changing the institutional position of the holder.

The custody boundary brings two property rights regimes into contact. Bitcoin's property rights regime embodies bearer instrument logic (Fairfield, 2015) because whoever controls the private key controls the asset without any intermediary required for transfer validity. Legacy financial governance assumes that property rights are mediated by identified custodians subject to jurisdictional authority (Diamond, 1984). At the custody interface, those logics support two coherent orientations or worldviews (see Chapter 8). Sovereignty-first and stability-first positions emerge from that boundary because each treats a different source of institutional assurance as primary. Sovereignty-first makes exit, direct control, and implementation level practices part of Bitcoin's institutional identity; stability-first makes regulated intermediation and fiduciary administrability the route through which Bitcoin can enter mainstream financial governance. The diagnostic problem is governance form fit across these competing accounts of what Bitcoin is for and what kind of agency its holders should retain.

The usual security-and-compliance discourse understates the governance challenge because it treats custody as a question of safe storage and secure access. Chapter 2's algorithmic velocity to institutional latency sequence locates the problem in the gap between fast operational movement and slower institutional judgment. Settlement, ETF flows, collateral use, and custodial reporting can move faster than fiduciary routines or protocol communities can interpret them. Informational turnover then changes the object of judgment because the institutional setting around custody keeps moving while actors are still deciding what custody dependence actually means.

Custody concentration is more legible than actionable. Disclosure can show when practical control has pooled inside regulated custody relationships but it rarely identifies an authority entitled to convert that evidence into a settlement over Bitcoin governance. Beneficial holders approach the boundary through fiduciary procedure, custodians through contract and regulation, and protocol participants through implementation norms and exit. Institutional latency appears at the point where those sources of authority must be made commensurable. Evidence of concentration can define the topic of concern but it cannot, by itself, create a governance form capable of reconciling action arenas and enabling binding governance choices.

That distinction separates the custody paradox from an ordinary financial concentration problem. Concentrated securities custodians, clearinghouses, and bank intermediaries can be addressed through supervisory law because the relevant authorities and regulated entities operate inside recognized legal hierarchies. Bitcoin custody concentration can receive some of the same treatment at the intermediary layer but the protocol layer remains outside that hierarchy. Regulation can discipline custodians without giving beneficial holders implementation level voice, and fiduciary process can make Bitcoin exposure administratively legible while reducing the holder's ability to exercise exit directly. Custody improves recourse inside legacy finance while weakening the governance agency that Bitcoin's bearer instrument design made possible.

I analyze that boundary as a structural governance problem. The four-level IAD framework (Ostrom and Ostrom, 2004; Rudd, 2004; Ostrom, 2005; Rudd, 2010), Williamsonian governance form selection (Williamson, 1985, 2002), Coasean transaction cost reasoning (Coase, 1937), and Hirschman's (1970) voice versus exit distinction identify what holders can actually do when the arrangement no longer fits their purposes. Together, those tools (Chapter 2) specify how custody converts economic exposure into asymmetric governance capacity. The question is not which worldview prevails but which institutional conditions allow the custody paradox to be managed.

Custody as institutional collision

Bitcoin's operational property rights have no direct equivalent in modern financial governance because ownership is exercised through the key rather than verified by a claim against an intermediary. A transaction is valid if the holder of the relevant private key signs it and the network's consensus mechanism accepts it (Nakamoto, 2008), without ratification by a custodian, court, or counterparty. Settlement is algorithmic, rapid, and irreversible. The private key is not evidence of ownership in the way a deed, share certificate, or bank account record is evidence; it is the operative mechanism through which ownership is exercised.

Legacy finance begins from the opposite institutional premise. Property rights in modern financial assets are claims that become usable because recognized intermediaries record, validate, and enforce them under jurisdictional authority (North, 1990; Pistor, 2019). Securities law builds this logic through custodial chains connecting beneficial owners to issuers (Schwarcz and Benjamin, 2002; Paech, 2016), while banking law embeds it in prudentially regulated deposit institutions (Diamond and Dybvig, 1983; Hellwig, 2009). In that setting, ownership is socially mediated before it is operationally effective.

The custody collision also raises questions about monetary purpose because legacy financial governance is organized around institutions that adjust quantities, supply liquidity, and absorb shocks through policy discretion (Friedman, 1968; Lavoie, 2014), whereas Bitcoin embeds a fixed supply rule enforced by protocol consensus. Demand shifts for Bitcoin drive price volatility (Rudd and Porter, 2025) because the elasticity in money supply that absorbs them in conventional systems is absent by design. Custody therefore brings an algorithmically-fixed monetary system into an institutional environment built around elastic money assumptions (Goodhart, 1989).

Note that Bitcoin's bearer instrument logic does not reject property rights (Fairfield, 2015) but implements them through a different institutional mechanism. Protocol consensus performs functions that custodial intermediation normally supplies and the custody boundary is where that institutional difference becomes practical.

Each IAD level contains a different version of the collision between Bitcoin's bearer rights logic and legacy finance's custodian-mediated logic (Table 5.1). The collision is muted at the operational level because the protocol does not distinguish between transactions authorized by individual keyholders and those authorized by institutional custodians. It is more consequential at the implementation level, where social norms determine whether a holder has practical standing. It becomes durable at the political and epistemic levels because the dispute has moved beyond key management, to the institutional commitments Bitcoin should preserve.

Table 5.1. The custody interface across Bitcoin’s IAD governance levels

The two custody forms have different governance impacts because they locate economically similar holders in different institutional positions. Self-custody holders can run nodes, select client implementations, and participate directly in implementation level governance (Antonopoulos, 2015). Institutional custody holders act through custodians whose operational decisions reflect regulatory obligations, fiduciary duties, and commercial incentives rather than protocol level representation. Identical economic exposure can thus translate into different governance agency solely because the custody arrangement differs.

Implementation level disputes over consensus rule changes, soft fork activation, and client software defaults depend on technical criteria and informal norms in the development community. Self-custody holders can enter that layer through node operation and software deployment, while institutional custody holders would need custodian-mediated channels that generally do not exist. The collision concentrates where contestation over what Bitcoin is for meets capital pools held by institutions who do not have voice in Bitcoin's own governance register.

The Coasean boundary and the mechanism of financialization

Coase’s (1937) analysis of the firm treats governance form as an economizing response to the cost of coordinating activity. The custody question is where and how control should be organized when Bitcoin moves from being an individual bearer asset into institutional portfolios. Self-custody leaves the holder responsible for the practical conditions of control, from key security and recovery, to software choice and exit. Institutional custody shifts that responsibility into a governed relationship that lowers some operational burdens by adding fiduciary process, regulatory recognition, and intermediary dependence. The custody boundary moves when the cost of either arrangement changes.

Transaction costs on each side of the boundary

The two sides of the custody boundary carry different transaction cost bundles (Table 5.2). Direct control concentrates operational competence and loss prevention on the holder, while institutional custody makes another organization responsible for control, reporting, and procedural assurance. The substitution is costly in a different register – counterparty exposure, compliance obligations, service agreements, and regulatory dependence become part of the price of using the institutional form.

Table 5.2. Transaction cost dimensions on each side of the custody boundary.

Boundary instability

The boundary is unstable because the transaction cost comparison changes with technology, regulation, holder type, and the scale of the Bticoin position. Improvements in wallets and collaborative custody can make direct control less costly, while ETF approval and qualified custodian pathways can make mediated control easier to justify inside fiduciary routines. The same custody form will therefore carry different institutional meaning for a retail holder, a family office, and a pension fund because competence, liability, approval, and reporting constraints do not fall on them in the same way. As the value of Bitcoin portfolios grows, the cost of error increases and the boundary can move again.

Trust in the surrounding institutional order also – crucially – enters the custody calculation. Institutional custody appears rational when courts function, enforcement is predictable, custodial neutrality is credible, and financial institutions remain trustworthy. When those conditions deteriorate, intermediary dependence becomes part of the risk being managed. Monetary instability, capital controls, banking failure, sanctions exposure, or rule-of-law deterioration can move the custody boundary without any change in Bitcoin's technical design.

Geographic variation in self-custody should be conceived in those terms, although available data do not yet identify custody type with enough granularity (Chainalysis, 2025). Jurisdictions facing currency collapse, capital controls, or sanctions exposure should exhibit stronger incentives toward self-custody than jurisdictions where intermediation remains trusted. The causal claim is this case is structural: the same Bitcoin technology sits inside different institutional environments but as macro structure and events unfold, the Coasean boundary continuously evolves.

Normative commitments also enter through the boundary. A holder with a strong sense of self-identity (Akerlof and Kranton, 2000), a Bitcoiners’ aspirational commitment to self-sovereignty for instance, may accept higher operational costs to stay on the self-custody side while a holder whose priority is preserving recourse may accept intermediation because it makes Bitcoin usable inside existing legal and fiduciary routines.

The Coasean boundary is therefore defined by the joint distribution of costs, structural conditions, and value filter commitments across the holder population. Movement across that boundary can be due to learning, regulatory change, institutional deterioration, or a change in what holders believe Bitcoin is for.

The boundary as the mechanism of financialization

Movement toward institutional custody changes Bitcoin's trajectory because it creates an installed base around which financialization actors can accumulate in advocacy coalitions or instrument constituencies (Sabatier and Weible, 2007; Voß and Simons, 2014). Once Bitcoin is held through standardized custodial relationships, financial actors build around the custody position and leave the protocol position mediated by another institution. The investment is institutional as much as technical: intermediaries learn how to document prudence, price exposure, organize collateral, and satisfy compliance routines around identifiable custodial relationships (Krippner, 2011; Davis and Kim, 2015). The custody boundary itself generates pathways forward because each successful custodial layer makes the next one easier to justify.

The spot Bitcoin ETF approvals accelerated that movement by changing the transaction costs facing institutional allocators (Langbein, 1995; SEC, 2024). They made Bitcoin exposure administratively compatible with existing portfolio governance while leaving key control outside the allocator's direct agency. Capital crossed the custody boundary because the approved vehicle made Bitcoin legible to fiduciary procedure. Once that vehicle existed, surrounding markets and services could organize around ETF-custodied Bitcoin, and custody concentration became part of the mechanism through which the new financial layer formed.

North's (1990) path dependence and Pierson's (1990) increasing returns logic explain why this movement cannot be treated as a reversible adoption choice. Compliance routines, fiduciary precedent, regulatory expectations, and operational integration teach actors to use the custodial form as the normal way Bitcoin enters institutional finance. Those investments narrow the feasible set of later configurations because alternatives must compete against an arrangement that has already accumulated professional legitimacy. Stability-first depends on this accumulated investment and its returns require a continuing population of custodial Bitcoin holders.

The same path dependence also identifies the conditions under which the stack can weaken. Gateway migration begins because a holder's individual experience under regulated exposure becomes an entry point into Bitcoin's bearer logic. Structural unwinding begins outside that learning process when monetary, legal, or financial institutions lose enough credibility that custodial dependence no longer appears prudent but becomes a non-trivial risk. In both cases, the Coasean boundary moves because the meaning of custody changes.

The fundamental transformation in custody

Williamson’s (1985) “fundamental transformation” appears once the custody relationship becomes costly to unwind. Before ETF approval, issuers could compare qualified custodians and preserve some bargaining leverage. After approval, filings, workflows, audit arrangements, market relationships, and fiduciary records begin to form around the selected custodian. The custody case differs from mining (Chapter 4) because the relationship-specific investment has no obvious conversion exit: the assets, procedures, and compliance routines are organized around Bitcoin custody itself.

The ETF approvals process

Before SEC approval of spot Bitcoin ETFs, the custody market for prospective issuers retained the discipline of ex ante competition. Multiple qualified custodians competed for mandates, and ETF applicants could evaluate regulatory status, operating capacity, insurance coverage, and service agreement terms before committing. In Williamson's (1985) sense, issuers could credibly switch among qualified custodians up until the point of filing.

After approval for ETF launches, Coinbase Custody won mandates from nine of the 11 spot ETF issuers and by April 2026 was responsible for approximately 84% of US spot Bitcoin ETF assets, worth roughly $77 billion in custodied Bitcoin. Coinbase entered the process with qualified custodian status, operational scale, compliance credibility, and the ability to meet a compressed launch timeline. Once the largest issuers selected that template, downstream service providers and boards evaluating later products had reason to treat Coinbase's position as evidence of prudence. First mover advantage became institutionalized through operational relationships (Arthur, 1989).

Concentration becomes analytically important when it reflects lock-in rather than temporary market share. Custodians configure operational infrastructure around ETF compliance, creation and redemption mechanics, audit routines, and fund reporting. Issuers build their own fiduciary record around the chosen custodian (Schaefer, 2019). Switching then requires regulatory refiling, operational transition, and market explanation. The relationship has moved from market selection to bilateral dependence.

Adaptive responses

Three Williamsonian (1985, 2002) responses are visible in the spot ETF ecosystem. Fidelity internalizes custody, while ARK 21Shares, CoinShares Valkyrie, and BlackRock spread risk across multiple custodians (Table 5.3). These arrangements reduce aggregate custodian concentration risk, but without removing bilateral dependence within each relationship.

Governance safeguards work through the terms of the custodial relationship itself. Contractual controls, audit requirements, insurance, key control procedures, and board oversight constrain custodian behavior after market discipline has weakened. In Williamson's terms, they are hybrid governance devices for managing the fundamental transformation rather than substitutes for it.

Table 5.3. Three Williamsonian responses to the fundamental transformation in ETF custody.

Asymmetry with mining

The mining case in Chapter 4 exhibits a related Williamsonian dynamic, with ex ante competition among host jurisdictions and energy providers followed by ex post bilateral dependence after jurisdiction-specific and contract-specific investments. Mining has a release valve that custody lacks because some infrastructure can be redirected toward AI-oriented HPC, preserving part of the capital investment while exiting the activity that generates the most Bitcoin-specific governance friction (Miles, 2018; Brown, 2025).

Bitcoin custody infrastructure has no comparable conversion exit. Its key management systems, compliance architecture, and protocol level operating knowledge are organized around Bitcoin custody and lose much – if not all – of their value outside that setting. Asset specificity is therefore even deeper than in mining. Vertical integration, governance safeguards, and co-custodian diversification remain responses inside Bitcoin custody rather than exits from the activity.

Other proof-of-stake networks also rely on concentrated qualified custody, so concentration by itself is not uniquely a Bitcoin phenomenon. Its meaning, however, changes from Bitcoin because Ethereum and Solana, for example, have foundation-mediated processes, validator-set governance rules, or stake-weighted channels through which large custodial positions acquire formal protocol voice. This brings another set of governance challenges (Fritsch et al., 2022; Jungnickel et al., 2025). For Bitcoin, institutional custody creates exposure without the standing that provides voice when protocol meaning is contested.

The structural paradox: largest stake, least governance agency

The fundamental transformation in custody produces Chapter 5's central paradox: participants with the largest economic stake in Bitcoin can occupy Bitcoin’s weakest governance position. The inversion follows from Bitcoin's permissionless IAD configuration (Chapter 3) because boundary rules do not limit participation based on economic stake and Bitcoin’s position rules do not privilege large holders. Custody in Bitcoin separates exposure value from governance standing.

Asset specificity is the mechanism that produces the inversion. Williamson (1985) treated specificity as a property imposed by transactional context but Bitcoin investors typically choose it by deliberately selecting institutional custody. That choice transforms their position from lower specificity, where exit remains cheap, to higher specificity, where exit becomes costly or procedurally difficult. Voice and exit are therefore partly endogenous to holders’ earlier custody choices; whether that inversion is a feature or a defect depends on epistemic values that institutional economics can help clarify but cannot settle.

The inversion is also constitutive of Bitcoin rather than an accidental governance defect. Permissionless participation – formalized during the early cypherpunk era (Jarvis, 2022; Nabben, 2023) – deliberately refused to make wealth, institutional status, or formal membership the basis for protocol authority. That refusal protects the system from a familiar form of capture, closing a pathway for concentrated capital to convert economic exposure into votes over consensus rules. Institutional custody changes the setting because large pools of capital enter Bitcoin through arrangements that would normally confer governance voice via recognizable fiduciary and administrative channels. The same design choice that protects Bitcoin from ownership-weighted control then leaves institutional holders without the voice they are accustomed to exercising in other asset classes.

The fiduciary constraint and three limitations

Institutional Bitcoin holders do not operate under the same conditions as retail participants because fiduciary duties require professional allocators to manage assets prudently for beneficiaries (Langbein, 1995). Those duties translate custody choices into procedures of due diligence, board authorization, compliance review, and legal assessment. Three limitations on institutional governance agency follow from the fiduciary constraint (Table 5.4).

Table 5.4. Three limitations on institutional governance agency arising from the fiduciary constraint

Institutional preferences often cannot register at the decision points where governance contests are settled because fiduciary procedures operate on slower review cycles than the technical and market events that establish facts on the ground. Bilateral dependencies weaken the threat of withdrawal that gives voice its force (Hirschman, 1970; Dowding et al., 2000). Institutional holders also do not generally occupy the informal implementation level channels through which Bitcoin governance is normally conducted. Their preferences can enter only indirectly and Bitcoin's governance architecture is not required to treat those indirect signals as binding.

Voice without a voice

Institutional holders cannot vote a proxy, file a shareholder resolution, or trigger a consultation requirement (McCahery et al., 2016) because Bitcoin has no corporate decision body to receive those acts. Voice remains institutionally undefined: a pension fund can speak, yet Bitcoin's permissionless architecture supplies no rule that converts the fund's economic stake into governance standing. The issue is political level standing rather than operational capacity. A custodian can sign transactions, secure keys, and report balances but those operational acts do not provide the custodian or the beneficial holder with any priveleged standing in protocol governance.

Institutional voice may reache Bitcoin through channels borrowed from legacy finance and public policy. Those channels can influence attention and may affect what developers, miners, exchanges, or custodians consider prudent, although Bitcoin's governance architecture assigns no weight to institutional preference as such. Recognition capacity can show that exposure is large and concentrated, and position papers, advocacy, or lobbying can make that evidence visible. They still do not create implementation level standing because custody locates holders outside the channels where informal norms are formed. Institutional latency appears when a recognized governance position cannot be converted into binding response in a timely manner.

Where governance leverage concentrates

Governance agency is concentrated less by wealth than by position inside Bitcoin's governance architecture. Developers, maintainers, and node operators – the technical minority – have implementation level standing because their leverage comes from technical credibility, social capital, and the operational capacity to catalyze or slow code deployment. The Core v30 / Knots dispute illustrates that concentration because the contest is conducted by actors operating within the informal norm layer that Chapter 3 identifies as Bitcoin's durable governance site. Its relevance for custody is not that every holder must, for example, have a view on relay policy; the dispute shows, however, where governance agency actually resides.Power is particularly concentrated amnog the large node runners who account for the economic majority of value transferred on the blockchain.

Self-custodial users supply a different source of governance leverage because key control makes exit credible. They can switch software implementations, withdraw support from contested positions, move assets across arrangements, or even relocate themselves jurisdictionally. A self-custody holder can move from Core to Knots by running different software, whereas an institutional holder cannot make the equivalent move without activating fiduciary processes that operate on slower and less flexible time frames.

Custodial holders were structurally absent from the Core v30 / Knots dispute even if some held strong views about relay policy. They lacked direct technical standing, operational flexibility, and credible exit, so a dispute that may affect them is being resolved through channels in which their economic stake does not create governance agency. The operational level can execute whatever valid transactions the prevailing software relays but the implementation, political, and epistemic levels determine which norms are treated as legitimate. Institutional custody locates large holders outside those slower levels even as it gives them exposure to the outcomes; their only way to influence decisions is through the informal norm layer, if they can align themselves with powerful node operators.

Bug or feature?

Whether the inversion of economic stake and governance agency is a feature or a defect depends on epistemic and political commitments that Bromley's (2006) volitional pragmatism treats as judgments about what there is reason to want. A sovereignty-first commitment treats the paradox as desirable because it prevents concentrated capital from capturing protocol governance, while a financialization-oriented commitment treats the same arrangement as a governance failure because the largest economic stakeholders have no formal agency over decisions that may affect the value they store. The same institutional fact supports different evaluations because the value filter differs.

My analysis leaves those commitments unresolved but specifies the fiduciary constraint, the self-imposed character of asset specificity, and the bifurcation of governance leverage that translate largest-stake positions into the weakest agency-oriented outcomes for institutional holders.

Conditions sustaining each orientation

Sovereignty-first and stability-first are sustained by different institutional investments. Sovereignty-first persists when holders maintain the practical capacity to verify, withdraw, run software, and bear the costs of direct control. Stability-first grows when regulated custody makes Bitcoin usable inside portfolio governance, compliance routines, and fiduciary precedent. The difference is institutional as well as ideological. Each orientation depends on a different stock of practices, organizations, and expectations that make one form of custody appear normal.

The two orientations occupy an uneven institutional field. Stability-first actors build organizations, compliance routines, and commercial expectations that continue to operate once created. Sovereignty-first actors preserve exit and direct control, yet those capacities become politically durable only when users, developers, advocates, and jurisdictions keep them visible. The asymmetry is temporal as well as organizational: even if the protocol remains formally permissionless, the surrounding institutional environment can become increasingly organized around custodial intermediation.

Constituency formation

Financialization generates organized constituencies (Sabatier and Weible, 2007; Pagliari and Young, 2014; Voß and Simons, 2014) because commercial actors acquire durable interests in regulated custodial Bitcoin exposure. ETF issuers, custodians, prime brokers, derivative counterparties, compliance firms, and legal advisers become participants in the political economy that later rule changes must confront. That organizational residue gives stability-first arrangements a constituency advantage over dispersed self-custody.

Self-custody lacks an equivalent constituency structure. A self-custodying individual may care deeply about sovereign control, yet that interest is dispersed, noncommercial, and difficult to mobilize. Advocacy organizations can defend the practice, although they do not command the resources available to financialization's institutional defenders. The constituency gap becomes consequential when recognition must become binding response: financialization has organized actors able to press for a regulated settlement, while self-custody often remains a capacity held separately by users.

Ratchet effects

Financialization also advances through ratchet effects. Custody regulation, fiduciary precedent, compliance infrastructure, reference pricing, derivatives, and collateral workflows become sunk investments (North, 1990; Pierson, 2000). Each investment is relationship-specific in Williamsonian (2002) terms and the effect extends beyond the custodian-client relationship. The financialization stack becomes harder to unwind because each layer teaches other actors to treat custodial Bitcoin as an ordinary institutional object.

Self-custody leaves a weaker institutional trace. It persists or erodes with technological accessibility, regulatory tolerance, and user competence, so a favorable political environment can turn hostile, even without accounting for the sunk-cost defenses that pro-financialization actors can bring to bear. A custodial product leaves behind legal templates, professional routines, and compliance expectations - institutional schemas (March and Olsen, 1984; Scott, 2005). An individual self-custody decision may preserve exit for that holder without creating comparable byproducts for later users.

Endogenous regulatory pressure

Regulatory pressure on self-custody is also endogenous to financialization's political economy. KYC and AML expansion at the protocol edge, capital gains treatment that burdens peer-to-peer settlement, and regimes that condition market access on custodial arrangements all grow from a framework organized around intermediated transactions (IRS, 2014; Nabilou, 2019; FATF, 2021; Baer et al., 2023; Barbereau and Bodó, 2023; EU, 2023; HM Treasury, 2023; Zetzsche et al., 2024). Non-intermediated transactions then appear as anomalies requiring justification.

The pressure also changes what counts as normal institutional conduct. Once regulated custody becomes the default pathway for fiduciaries, non-intermediated control begins to look like an exception requiring explanation. That shift reaches prudence, regulatory accommodation, and standing when disputes over Bitcoin's future arise because it reflects political and epistemic values as well as operational processes.

Counterweights to path asymmetry

The path-asymmetric bias toward stability-first has counterweights, although they work through different institutional channels (Table 5.5). Gateway migration changes the holder's understanding of custody from inside the experience of regulated exposure. Structural unwinding changes the background conditions that made custodial dependence appear prudent. A hard fork would work through separation rather than adjustment, producing protocol trajectories for commitments that could no longer coexist inside Bitcoin.

Table 5.5. Three mechanisms operating against path-asymmetric bias toward stability-first equilibria

Gateway mechanics

Gateway migration – sometimes referred to as Bitcoin’s Trojan Horse scenario – occurs when holders who enter through regulated vehicles or corporate treasury arrangements later move toward self-custody. Exposure can precede institutional learning: a holder may first treat custody as convenience and later see bearer control, node operation, or exit as the features that make Bitcoin distinctive. Custodial disruption, regulatory constraint, larger balances, or governance disputes can reveal risks that the initial custody choice underweighted. The gateway expands the sovereignty-first population through experience rather than through a lobbying apparatus that stability-first has at its disposal.

Structural unwinding

Structural unwinding operates from outside Bitcoin's custody system because stability-first remains rational only while monetary, legal, regulatory, and intermediary conditions remain trustworthy (Reinhart and Rogoff, 2009; Eichengreen, 2011). A currency crisis, banking failure, capital control episode, or breakdown in custodial neutrality can make self-custody the lower cost option without any protocol change or ideological conversion. The Coasean boundary moves because the world around the holder changes.

The empirical uncertainty differs across the two pathways. Gateway migration requires observation across cohorts and custody pathways; structural unwinding depends on macroeconomic, geopolitical, and government failure conditions outside Bitcoin's own design. Each pathway leaves different evidence about how holders respond when custody dependence changes meaning.

The hard-fork option

A hard fork (Bier, 2021) would incorporate features stability-first requires and Bitcoin resists: formal institutional voice; reversible transactions under defined conditions; KYC-compatible address structures; and identifiable transfer authority. A hard fork would separate the two worldviews into distinct protocols instead of reconciling them inside Bitcoin.

Although institutionally available, a hard fork would be politically costly. It would require coordination among actors whose relationships to Bitcoin are organized through different governance forms, making the coordination costs unusually high for a permissionless system. Each individual holder would face a discrete choice between continuity with the original protocol or access to a financialized variant. Whether such a fork would preserve network effects remains speculative.

The trajectory question

The trajectory question is whether the window in which both orientations remain live options is narrowing. Table 5.6 organizes the relevant conditions across domains that move at different institutional speeds. Path asymmetry biases the regulatory, jurisdictional, and institutional domains toward stability-first. Gateway migration works through technological and experiential change, while structural unwinding works through the background conditions that make custodial intermediation rational.

Table 5.6. Conditions sustaining the viability of each institutional logic

The domains in Table 5.6 move on different clocks. Wallet design and collaborative custody can preserve self-custody accessibility, although they must keep pace with the professionalization of regulated custody. Regulatory and jurisdictional developments can entrench one side faster than user practice adapts, especially where market access is conditioned on intermediated channels. Fiduciary precedent and compliance normalization move more slowly but become harder to reverse once embedded in professional routines. The trajectory is therefore a multi-level institutional adjustment process in which fast operational adoption may precede slower political and epistemic settlement.

Reversing path asymmetry under stable conditions would require investment in self-custody technology, regulatory defense, jurisdictional support, and institutional advocacy. Structural unwinding follows a different route because it changes the conditions that made custody rational rather than requiring comparable mobilization by self-custody holders. Whether financialization should be resisted, accommodated, or accelerated remains a volitional question in Bromley's (2006) sense: it depends on what Bitcoin is for, which dependencies are acceptable, and which forms of agency should be preserved.

Information gaps and research needs

Research on custody should begin from the boundary condition. Aggregate concentration records where assets sit; it does not show whether custody has become an institutional position that changes the relation among exposure, control, standing, and exit. Concentration becomes diagnostic when relationship-specific investments make exit costly and give intermediaries durable positions in the financialization stack. ETF flows are evidence of institutionalization only when regulated exposure becomes the ordinary route through which fiduciaries encounter Bitcoin. Regulatory treatment adds another test: whether that route is gaining enough organizational force to make self-custody appear exceptional even where direct control remains technically available. The counterevidence is the continued practical use of direct control where trust in monetary, legal, regulatory, or financial institutions weakens (see Chapter 8).

Evidence should therefore be read for the authority claims embedded in ordinary records. Custody agreements allocate action under stress; who can move assets, pause withdrawals, respond to forks, and translate protocol uncertainty into fiduciary procedure? Board records and fiduciary policies show how Bitcoin is recast as prudence, liquidity, operational risk, or compliance exposure. Regulatory filings show whether custodial dependence is being treated as ordinary infrastructure or a governance exposure requiring justification. Governance-event reconstruction then asks whether actors who could recognize a custody problem had standing in the arena where response was possible. That kind of evidence would help clarify where recognition, standing, and exit separate, and whether that separation settles Bitcoin's competing purposes before they have been explicitly argued.

Conclusion

Bitcoin's custody paradox is not a puzzle about misplaced preferences. The same movement that makes Bitcoin easier to hold inside regulated instruments can weaken the agency that bearer ownership was supposed to preserve. Coase's boundary logic explains why holders move between direct and intermediated control as transaction costs change, while Williamson's fundamental transformation shows why ETF custody can harden into ex post dependence once fiduciary, operational, and compliance investments accumulate. Hirschman's voice and exit logic then identifies the governance consequence: exposure can grow at the same time that standing and exit become harder to exercise. Custody solves an access problem by creating a governance problem.

Path-asymmetric financialization remains contingent because its counterweights operate under different institutional conditions. Gateway migration shifts holders toward self-custody when regulated exposure produces learning about Bitcoin's bearer logic instead of durable dependence on intermediaries. Structural unwinding, however, shifts the boundary when background institutions lose enough credibility that custody no longer appears prudent. Hard fork separation would be a more explicit institutional settlement because it would sort stability-first and sovereignty-first commitments into different protocol trajectories. The volitional question is which dependencies actors have reason to accept, and which governance capacities those dependencies leave available when Bitcoin's purposes are contested.

Custody is a site of institutional latency because recognition of concentration does not create a legitimate binding response. Better disclosure can identify where control has pooled, and recognition capacity can make the custody signal more visible. Neither, however, supplies standing at Bitcoin's implementation level. Informational turnover in financial products and regulatory expectations can further obscure the signal by changing the object being measured while institutions are still interpreting it. Governance form mismatch appears when the arrangement that lowers immediate transaction costs also narrows the future capacity to decide what Bitcoin is for.

The custody paradox is therefore not about concentration alone. A more transparent custody market, a better regulated custodian, or a more liquid ETF structure can still generate governance friction if institutional custody becomes the normal route into Bitcoin while direct control loses practical salience. Regulated exposure may become the default form through which many holders encounter Bitcoin if institutional custody continues to reduce access costs and financialization builds stronger constituencies around that form. Sovereignty-first remains viable where gateway migration, self-custody technology, legal protection, or structural unwinding keep exit practical. Which configuration preserves enough recognition, standing, and exit to keep Bitcoin's competing purposes contestable before custody arrangements settle them by default remains an open question.

References

Akerlof, GA and RE Kranton 2000. Economics and identity. The Quarterly Journal of Economics 115: 715-753. http://www.jstor.org/stable/2586894

Antonopoulos, AM 2015. Mastering Bitcoin: Unlocking Digital Cryptocurrencies. Sebastol, CA: O'Reilly Media, Inc.

Arthur, WB 1989. Competing technologies, increasing returns, and lock-in by historical events. The Economic Journal 99: 116-131. https://doi.org/10.2307/2234208

Baer, K, R De Mooij, S Hebous, et al. 2023. Taxing cryptocurrencies. Oxford Review of Economic Policy 39: 478-497. https://doi.org/10.1093/oxrep/grad035

Barbereau, T and B Bodó 2023. Beyond financial regulation of crypto-asset wallet software: in search of secondary liability. Computer Law & Security Review 49: 105829. https://doi.org/10.1016/j.clsr.2023.105829

Bier, J 2021. The Blocksize War: The Battle Over Who Controls Bitcoin’s Protocol Rules: Independently published.

Bromley, DW 2006. Sufficient Reason: Volitional Pragmatism and the Meaning of Economic Institutions. Princeton NJ: Princeton University Press.

Brown, G 2025. Reclaiming Public Infrastructure: Public Power New York and the Struggle for Energy Democracy. PhD dissertation, University of Glasgow, https://theses.gla.ac.uk/85234/

Chainalysis 2025. The 2025 geography of crypto report. https://www.chainalysis.com/wp-content/uploads/2025/10/the-2025-geography-of-crypto-report-release.pdf

Coase, RH 1937. The nature of the firm. Economica 4: 386–405. https://doi.org/10.2307/2626876

Davis, GF and S Kim 2015. Financialization of the economy. Annual Review of Sociology 41: 203-221. https://doi.org/10.1146/annurev-soc-073014-112402

Diamond, DW and PH Dybvig 1983. Bank runs, deposit insurance, and liquidity. Journal of Political Economy 91: 401-419. http://www.jstor.org/stable/1837095

Diamond, DW 1984. Financial intermediation and delegated monitoring. The Review of Economic Studies 51: 393-414. https://doi.org/10.2307/2297430

Dowding, K, P John, T Mergoupis, et al. 2000. Exit, voice and loyalty: analytic and empirical developments. European Journal of Political Research 37: 469-495. https://doi.org/10.1111/1475-6765.00522

Eichengreen, B 2011. Exorbitant Privilege: The Rise and Fall of the Dollar and the Future of the International Monetary System. Oxford, UK: Oxford University Press.

EU 2023. Regulation (EU) 2023/1114 of the European Parliament and of the Council of 31 May 2023 on markets in crypto-assets. https://eur-lex.europa.eu/eli/reg/2023/1114/oj/eng

Fairfield, J 2015. BitProperty Southern California Law Review 88: 805-874. https://southerncalifornialawreview.com/2015/05/02/bitproperty-article-by-joshua-a-t-fairfield/

FATF 2021. Updated guidance for a risk-based approach to virtual assets and virtual asset service providers. www.fatf-gafi.org/publications/fatfrecommendations/documents/Updated-Guidance-RBA-VA-VASP.htm,

Friedman, M 1968. The role of monetary policy. American Economic Review 58: 1-17. https://www.aeaweb.org/aer/top20/58.1.1-17.pdf

Fritsch, R, M Muller and R Wattenhofer 2022. Analyzing voting power in decentralized governance: who controls DAOs? arXiv preprint, https://arxiv.org/pdf/2204.01176

Goodhart, CAE 1989. Money, Information and Uncertainty, Second Edition. Cambridge, MA: The MIT Press.

Hellwig, MF 2009. Systemic risk in the financial sector: an analysis of the subprime-mortgage financial crisis. De Economist 157: 129-207. https://doi.org/10.1007/s10645-009-9110-0

Hirschman, AO 1970. Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States. Cambridge, MA: Harvard University Press.

HM Treasury 2023. Future financial services regulatory regime for cryptoassets: consultation: response to the consultation and call for evidence. https://assets.publishing.service.gov.uk/media/653bd1a180884d0013f71cca/Future_financial_services_regulatory_regime_for_cryptoassets_RESPONSE.pdf

IRS 2014. Notice 2014-21. https://www.irs.gov/pub/irs-drop/n-14-21.pdf

Jarvis, C 2022. Cypherpunk ideology: objectives, profiles, and influences (1992–1998). Internet Histories 6: 315-342. https://doi.org/10.1080/24701475.2021.1935547

Jungnickel, M, FÖ Sönmez, C Mulligan, et al. 2025. DAO governance: voting power, participation, and controversy - a review and an empirical analysis. Distributed Ledger Technologies: Research and Practice. https://doi.org/10.1145/3777416

Krippner, GR 2011. Capitalizing on Crisis: The Political Origins of the Rise of Finance. Cambridge, MA: Harvard University Press.

Langbein, JH 1995. The contractarian basis of the law of trusts. The Yale Law Journal 105: 625-675. https://doi.org/10.2307/797196

Lavoie, M 2014. Post-Keynesian Economics: New Foundations. Cheltenham, UK: Edward Elgar.

March, JG and JP Olsen 1984. The New Institutionalism: organizational factors in political life. The American Political Science Review 78: 734-749. https://doi.org/10.2307/1961840

McCahery, JA, Z Sautner and LT Starks 2016. Behind the scenes: the corporate governance preferences of institutional investors. The Journal of Finance 71: 2905-2932. https://doi.org/10.1111/jofi.12393

Miles, K 2018. Bitcoin is eating Quebec. MIT Technology Review, April 11, 2018, https://www.technologyreview.com/2018/04/11/143987/bitcoin-is-eating-quebec/

Nabben, K 2023. Cryptoeconomics as governance: an intellectual history from “Crypto Anarchy” to “Cryptoeconomics”. Internet Histories 7: 254–276. https://doi.org/10.1080/24701475.2023.2183643

Nabilou, H 2019. How to regulate Bitcoin? Decentralized regulation for a decentralized cryptocurrency. International Journal of Law and Information Technology 27: 266-291. https://doi.org/10.1093/ijlit/eaz008

Nakamoto, S 2008. Bitcoin: a peer-to-peer electronic cash system. Unpublished white paper, https://bitcoin.org/bitcoin.pdf

North, DC 1990. Institutions, Institutional Change and Economic Performance. Cambridge MA: Cambridge University Press.

Ostrom, E and V Ostrom 2004. The quest for meaning in public choice. American Journal of Economics and Sociology 63: 105-147. https://doi.org/10.1111/j.1536-7150.2004.00277.x

Ostrom, E 2005. Understanding Institutional Diversity. Princeton, N.J.: Princeton University Press.

Paech, P 2016. Securities, intermediation and the blockchain: an inevitable choice between liquidity and legal certainty? Uniform Law Review 21: 612-639. https://doi.org/10.1093/ulr/unw040

Pagliari, S and KL Young 2014. Leveraged interests: financial industry power and the role of private sector coalitions. Review of International Political Economy 21: 575-610. https://doi.org/10.1080/09692290.2013.819811

Pierson, P 2000. Increasing returns, path dependence, and the study of politics. American Political Science Review 94: 251-267. https://doi.org/10.2307/2586011

Pistor, K 2019. The Code of Capital: How the Law Creates Wealth and Inequality. Princeton, N.J.: Princeton University Press.

Reinhart, CM and KS Rogoff 2009. This Time Is Different: Eight Centuries of Financial Folly. Princeton, N.J.: University Press.

Rudd, MA 2004. An institutional framework for designing and monitoring ecosystem-based fisheries management policy experiments. Ecological Economics 48: 109-124. https://doi.org/10.1016/j.ecolecon.2003.10.002

Rudd, MA 2010. A logic model for assessing the sustainability of Canadian oceans policy and management. In Ocean Yearbook 24, eds. A Chircop, S Coffen-Smout and M McConnell, 9-36. Leiden, The Netherlands: Martinus Nijhoff Publishers. http://www.brill.com/ocean-yearbook-24

Rudd, MA and D Porter 2025. Bitcoin supply, demand, and price dynamics. Journal of Risk and Financial Management 18: 570. https://doi.org/10.3390/jrfm18100570

Sabatier, S and CM Weible 2007. The Advocacy Coalition Framework. In Theories of the policy process, ed. S Sabatier, 189-220. Cambridge MA: Westview Press. https://doi.org/10.4324/9780367274689

Schaefer, DC 2019. Applying the SEC Custody Rule to cryptocurrency hedge fund managers. California Law Review 107: 1381-1414. https://doi.org/10.15779/Z38M03XX77

Schwarcz, SL and J Benjamin 2002. Intermediary risk in the indirect holding system for securities. Duke Journal of Comparative & International Law 12: 309-330. https://scholarship.law.duke.edu/djcil/vol12/iss2/3

Scott, WR 2005. Institutional theory contributing to atheoretical research program. In Great Minds in Management: The Process of Theory Development, eds. KG Smith and MA Hitt, 460–484. Oxford UK: Oxford University Press. https://doi.org/10.1093/oso/9780199276813.003.0022

SEC 2024. Order granting accelerated approval of proposed rule changes, as modified by amendments thereto, to list and trade Bitcoin-based Commodity-Based Trust Shares and Trust Units. Release No. 34-99306, https://www.sec.gov/files/rules/sro/nysearca/2024/34-99306.pdf

Voß, J-P and A Simons 2014. Instrument constituencies and the supply side of policy innovation: the social life of emissions trading. Environmental Politics 23: 735-754. https://doi.org/10.1080/09644016.2014.923625

Williamson, OE 1985. The Economic Institutions of Capitalism. New York: The Free Press.

Williamson, OE 2002. The theory of firm as governance structure: from choice to contract. The Journal of Economic Perspectives 16: 171–195. http://www.jstor.org/stable/3216956

Zetzsche, D, J Sinnig and A Nikolakopoulou 2024. Crypto custody. Capital Markets Law Journal 19: 207-229. https://doi.org/10.1093/cmlj/kmae010