The Cost of Permission
Permissionless systems let anyone in. They do not let anyone decide.
Permissionless systems let anyone in. They do not let anyone decide. Someone pays for the difference — and nobody has counted how much.
Keywords: permissionless blockchains · protocol governance · hold-up · asset specificity · switching costs · complementors · the Merge · Ethereum · institutional economics · industrial organisation
This is an idea, not the paper. The empirical work described below is at the probe stage. Where I have checked something myself, I say so. Where I have not, I say that too.
Open admission is a claim about entry. It is not a claim about governance.
Anyone may run a node. Anyone may deploy a contract. No gatekeeper, no application, no licence. That door is genuinely open, and the people who built these systems were right to be proud of it.
The second door is different. Rule changes originate with a client-development coalition, get ratified by validators and exchanges, and arrive at everyone else as a fact. The firms that built on the protocol are consulted. They do not decide. When the rules change, they rebuild or they leave.
Figure 1. Entry and rule change are separate questions. Conflating them is where the word “permissionless” does its damage.
I made this argument at length in the Journal of Institutional Economics earlier this year — that open entry and governance over rule change are distinct properties, and that rule-changing coalitions can revise the protocol after complementors have sunk chain-specific investment (Wright, 2026). The mechanism is ordinary. It is hold-up: specific investment, plus residual control held by somebody else, plus an ex post change in terms (Klein, Crawford & Alchian, 1978; Williamson, 1979; Grossman & Hart, 1986).
What is not ordinary is the setting. These systems were designed specifically to remove the party who could do this. The whitepapers are explicit about it. And yet governance concentrated in client teams, miners and stakers reconstitutes exactly the residual claimant the design was meant to eliminate. The problem was not solved. It was relocated.
That paper made the argument. It did not answer the question that follows, and the question that follows is the one that matters to anybody outside the seminar room:
What did it cost?
The number nobody has
Consider the Ethereum Merge, 15 September 2022. Proof-of-work to proof-of-stake, executed cleanly, widely praised as an engineering achievement. It was one. It also obsoleted an entire category of capital equipment and an entire category of firm overnight, on a schedule those firms did not set.
That is an architectural change in the sense of Henderson and Clark (1990) — the components survive, the way they fit together does not, and incumbent competence evaporates. Iansiti and Khanna (1995) documented the same pattern in firm capabilities. The theory is in place. The measurement is not.
So I went looking for the number. Two probes, both run this month.
Probe one: audited filings. The SEC’s full-text search and XBRL endpoints are public and I queried them directly. The population of Ethereum-mining-exposed registrants is roughly a dozen, and most of them are not 10-K filers. HIVE files 40-F. Bit Digital, BIT Mining, SOS and Nano Labs file 20-F. HIVE reports under IFRS; Bit Digital, Core Scientific and Sysorex report under US GAAP — I pulled the company-facts records and confirmed the taxonomies differ. Core Scientific, the one with the most complete impairment tagging, was predominantly a Bitcoin miner and filed Chapter 11 three months after the Merge, so its writedowns say nothing useful about proof-of-stake.
And here is the harder problem. An impairment charge is not an adaptation cost. The 2022 writedowns in this sector confound a crypto price collapse, an energy price shock, a financing squeeze and the Merge, all at once. Reading a protocol-change cost out of an impairment line is not measurement. It is inference dressed as observation.
Probe two: the developer record. GH Archive publishes the entire public GitHub event stream. I pulled the hour beginning 12:00 UTC on Merge day — the transition finalised around 06:42 that morning — and counted.
Figure 2. Left: 206,070 events globally in that hour; 175 on Ethereum client and tooling repositories. Right: where that activity sat. Counted directly from the GH Archive hourly file, 27 July 2026.
Two things fall out of this. The first is scale: one hour is 101 MB compressed, about 2.4 GB a day, so a six-month window is roughly 440 GB. That is a BigQuery job, not a laptop.
The second is more interesting, and it is the reason I am writing this as an idea rather than a result. My keyword filter matched two Cloud Foundry repositories, because “cloudfoundry” contains “foundry.” That is funny for about four seconds and then it is the whole problem. Defining which repositories constitute firms building on Ethereum is not a free step. GitHub gives you repository names and account logins, not firm identity. And a push event is not a wage bill. Activity is not effort, and effort is not cost.
Why this is an industrial organisation problem
The closest thing to a template is nearly thirty years old and sits in the journal I am aiming at. Greenstein (1997) measured lock-in and switching costs among mainframe buyers by asking what the buyers themselves saw. Not what the vendor did — what it cost the people on the other end. That is the correct posture, and I do not think anyone has adopted it for protocol change.
The adjacent literature is closer than it looks. Farrell and Saloner (1985, 1988) on standardisation and coordination through committees describes protocol governance with unnerving accuracy. Boudreau (2010) on granting access versus devolving control is the distinction I am drawing, in a different vocabulary. Eisenmann, Parker and Van Alstyne (2011) on platform envelopment covers the case where the platform moves and complementors absorb it. Argyres, Nickerson and Ozalp (2023) put adjustment, transaction and opportunity costs on complementor responses directly.
And the crypto-economics literature has been circling the governance problem without quite landing on the cost question. Budish (2025) establishes economic limits on trust at scale. Biais, Bisière, Bouvard and Casamatta (2019) show that consensus supports multiple equilibria — forks are not accidents, they are on the menu. Davidson, De Filippi and Potts (2018) and Alston, Law, Murtazashvili and Weiss (2022) treat these systems as institutions. None of them counts what a rule change costs the firms downstream.
What I will not claim
I do not have the number. I have two data sources that each capture something adjacent to it and neither of which is a cost. Joining them requires a mapping from repositories to firms that does not yet exist, and validating that mapping requires a subsample where the true cost is observable — which, on current evidence, is about four companies under two accounting standards.
That is the state of it. The mechanism is established. The magnitude is open. Anyone who tells you they know what the Merge cost the ecosystem is guessing, and so far that includes me.
If you ran a business through a protocol transition — the Merge, SegWit, Taproot, a chain split, any of it — and you know what it cost you in engineering months or written-off hardware, I would like to hear from you. That is the missing piece, and it does not exist in any database I can reach.
References
Alston, E., Law, W., Murtazashvili, I., & Weiss, M. (2022). Blockchain networks as constitutional and competitive polycentric orders. Journal of Institutional Economics, 18(5), 707–723. https://doi.org/10.1017/S174413742100093X
Argyres, N., Nickerson, J., & Ozalp, H. (2023). Platform competition and complementor responses. Industrial and Corporate Change, 32(1), 112–128. https://doi.org/10.1093/icc/dtac027
Biais, B., Bisière, C., Bouvard, M., & Casamatta, C. (2019). The blockchain folk theorem. The Review of Financial Studies, 32(5), 1662–1715. https://doi.org/10.1093/rfs/hhy095
Boudreau, K. (2010). Open platform strategies and innovation: Granting access vs. devolving control. Management Science, 56(10), 1849–1872. https://doi.org/10.1287/mnsc.1100.1215
Budish, E. (2025). Trust at scale: The economic limits of cryptocurrencies and blockchains. The Quarterly Journal of Economics, 140(1), 1–62. https://doi.org/10.1093/qje/qjae033
Davidson, S., De Filippi, P., & Potts, J. (2018). Blockchains and the economic institutions of capitalism. Journal of Institutional Economics, 14(4), 639–658. https://doi.org/10.1017/S1744137417000200
Eisenmann, T., Parker, G., & Van Alstyne, M. (2011). Platform envelopment. Strategic Management Journal, 32(12), 1270–1285. https://doi.org/10.1002/smj.935
Farrell, J., & Saloner, G. (1985). Standardization, compatibility, and innovation. The RAND Journal of Economics, 16(1), 70. https://doi.org/10.2307/2555589
Farrell, J., & Saloner, G. (1988). Coordination through committees and markets. The RAND Journal of Economics, 19(2), 235. https://doi.org/10.2307/2555702
Greenstein, S. M. (1997). Lock-in and the costs of switching mainframe computer vendors: What do buyers see? Industrial and Corporate Change, 6(2), 247–273. https://doi.org/10.1093/icc/6.2.247
Grossman, S. J., & Hart, O. D. (1986). The costs and benefits of ownership. Journal of Political Economy, 94(4), 691–719. https://doi.org/10.1086/261404
Henderson, R. M., & Clark, K. B. (1990). Architectural innovation. Administrative Science Quarterly, 35(1), 9. https://doi.org/10.2307/2393549
Iansiti, M., & Khanna, T. (1995). Technological evolution, system architecture and the obsolescence of firm capabilities. Industrial and Corporate Change, 4(2), 333–361. https://doi.org/10.1093/icc/4.2.333
Klein, B., Crawford, R. G., & Alchian, A. A. (1978). Vertical integration, appropriable rents, and the competitive contracting process. The Journal of Law and Economics, 21(2), 297–326. https://doi.org/10.1086/466922
Williamson, O. E. (1979). Transaction-cost economics: The governance of contractual relations. The Journal of Law and Economics, 22(2), 233–261. https://doi.org/10.1086/466942
Wright, C. S. (2026). Beyond ‘permissionless’: Governance, commitment, and rule change in public blockchains. Journal of Institutional Economics, 22, e21. https://doi.org/10.1017/S1744137426100599