A disclosure first, restated in full because this is the article where it matters. I am the founder of a company that offers an employer-sponsored Bitcoin savings benefit. You should weigh that against everything below. I am going to run the asset through a framework I published before I applied it here, without altering a word of it, and it is going to fail several of the tests outright. That is the point of building the test first.
In October 2008 a pseudonymous author published a nine-page paper describing a peer-to-peer electronic cash system, and in January 2009 the system began operating.
It drew on a lineage of prior work: timestamped chains of records, proof-of-work as a costly signal, earlier proposals for decentralized digital cash and scarce digital objects, and the older computer-science problem of reaching agreement among mutually distrustful parties. What was new was the combination of a ledger maintained by open participation rather than a trusted institution, secured by computational work, with a supply schedule fixed in the protocol.
Only the properties that bear on this argument matter here.
Issuance is predetermined. New units enter circulation on a schedule written into the protocol, at a rate that halves roughly every four years, converging on a total that will never exceed twenty-one million units, most of which have already been issued. There is no issuer in the ordinary sense, and changing the schedule would require coordinated agreement of the network's participants against their own economic interest.
Ownership is established by cryptographic keys and transfer by a signed transaction the network validates, with settlement final once sufficiently confirmed and no clearing institution involved.
Custody is a genuine choice. Units can be held directly by the owner who holds the keys, or through an intermediary, in which case the owner holds a claim on the intermediary rather than the units. The second paper's distinction between property at the value layer and promise at the custody layer applies here with full force.
The asset is portable in a way physical assets are not, and a holder who knows their keys can carry any quantity across any border, subject to the laws of both jurisdictions.
And it is volatile. Its price in national currencies has moved with an annualized volatility several times that of broad equity indices and gold, with repeated peak-to-trough declines exceeding half its value, including a decline of roughly three-quarters across 2022.
Stated neutrally: a monetary experiment with a materially different architecture from national currencies, rule-governed in supply, without an issuer at the value layer, portable across institutions and jurisdictions, and carrying in exchange severe volatility, a short history, and technological, custodial, regulatory and behavioral risks that national currencies do not.
An Asset Is Not a Standard
Before any test is applied, one distinction has to be made explicit, because most of the public argument about this asset founders on its absence.
There are two propositions in circulation, with completely different evidentiary burdens.
The first is that the asset can serve as one component of an individual's or a plan's preservation architecture, held alongside productive assets and promises, for a particular function within a composed whole.
The second is that an economy could or should operate on a Bitcoin monetary standard, with the asset as the unit of account and medium of exchange for wages, prices, debts and taxes.
The second is what the historical and economic criticism of rigid money is about, and that criticism is formidable. A fixed-supply money in a growing economy implies a falling price level. Falling prices under nominal debt produce debt-deflation. A money that cannot be created cannot be lent in a panic. Nominal wages that resist reduction would, under persistent deflation, produce unemployment through exactly the channel that defends positive inflation on compensation's own terms. Keynes, Friedman and the interwar record all bear against a Bitcoin standard, and nothing in this paper answers any of it.
I do not require the second proposition to be true for the first to be useful, and I decline to defend it.
Whether an individual worker, inside an economy that continues to operate on discretionary national money, might hold some portion of deferred compensation in a form of monetary property with a rule-governed supply is separable from whether the whole economy should adopt the asset as its unit. A worker who holds gold is not proposing a return to 1913.
Importing conclusions from the standard question into the component question, in either direction, is a category error. Critics and advocates who do it are making the same mistake from opposite sides.
The Tests, Unaltered
The Six Tests are applied exactly as written. No test added, removed, reworded or reweighted. The specified worker is the ordinary formal-economy employee this trilogy has addressed throughout: forty-year horizon at the start of the career, a liability tranched into floor, flexible and aspirational portions, no professional expertise, an employer plan as the primary vehicle, and a career that will cross several employers and possibly a border.
Arrival asks whether the architecture credibly carries value to the horizon at which the liability falls due, measured against the liability. Volatility governs the answer. For the floor tranche, the non-negotiable minimum that tolerates almost no shortfall, the asset fails plainly. An asset that has lost more than half its currency value on several occasions within a decade and a half, and roughly three-quarters in the most recent such episode, cannot credibly carry a floor liability to a fixed date. A worker whose rent in retirement depended on its price at the moment of retirement would face sequence risk in its most concentrated form. For the flexible and aspirational tranches the test is unresolved rather than failed: the history is too short and the regime dependence too little understood to establish liability-relative reliability across forty years. The empirical literature on the asset as an inflation hedge is mixed, with some studies finding the property in some periods and others finding none outside particular episodes. Unresolved means unresolved in both directions. It is not a soft pass and it is not a soft fail.
Assignment asks which risks the architecture reduces, transfers, transforms or creates, and who bears them. This is the test on which the asset's case principally rests. It removes, or sharply constrains, one particular source of monetary uncertainty: discretionary alteration of the unit's supply schedule by an authority whose incentives may diverge from the holder's. That constraint is real and structural. But it does not remove purchasing-power uncertainty. It removes one source and leaves the rest untouched: productivity, demand, fiscal policy, supply shocks, relative prices, capital flows, and the market's own valuation of the asset. What the constraint buys is a shorter list of sources. What it costs is a change in the character of the uncertainty that remains, from relatively gradual and roughly predictable unit decay to much greater and less predictable market-price variance. And it creates risks the currency holder does not bear: custody risk, technological risk, regulatory risk, tax risk, and behavioral risk in amplified form, since the volatility that disqualifies it for the floor is exactly the volatility that induces panic selling at troughs. The honest result is that it reassigns risk rather than reducing it on net. It removes the issuer and hands the holder the market, the keys, the technology, the regulator, and their own psychology.
Ownership asks what the worker actually holds, at the value layer and the custody layer separately. At the value layer this is an unusual object: not a promise in the ordinary sense, since no counterparty owes the holder anything, but not property in the sense a bar of gold is, since its value depends on the continued operation of a network and the willingness of others to accept it. The best description I can offer is property without an issuer, nearer the property pole than any financial asset and nearer the promise pole than any physical one. At the custody layer, everything depends on the form of holding. Directly held, with the worker controlling the keys, it is property at both layers. Custodial holdings, exchange balances and fund shares are promises at the custody layer, and the history of intermediary failures in this asset is recent and severe enough that the point needs no elaboration.
Journey asks whether the architecture survives the career. On portability of the claim, it performs as well as anything in the inventory: it never belonged to the employer, it is not anchored to a provider, and it crosses jurisdictions as information. On portability of the architecture, the second paper's more consequential distinction, it performs no better than any other asset, because the architecture is not the asset. A holding inside an employer plan carries the plan's defaults, automation, tax treatment and governance, and none of that travels when the worker leaves. Portability across jurisdictions is not independence from jurisdictions. And on temporal survival, fifteen years is a shorter track record than any other instrument in the inventory, which makes this unresolved.
Human asks whether an ordinary person, busy with a life, can begin, persist with and convert this architecture, and it scores required action as a defect. Held directly, with the worker managing keys, it fails for most actual humans. Self-custody requires generating, securing, backing up and never losing cryptographic keys, with no recourse for error, and the documented history of lost keys, mistaken transfers and phishing losses is the test failing in real time. Volatility compounds it. And initiation requires the worker to seek it out, understand it and act, which is the opposite of an inaction path.
Conversion asks whether accumulated value can become the living it was preserved for, and whether a path to pooling exists where the liability is longevity-contingent. The asset provides no pooling. It carries no mortality credits, insures no longevity, and cannot become an income that survives its holder's lifespan without being converted into a promise from an institution that can pool. Conversion into consumption requires sale into currency at whatever price prevails at the horizon, so sequence risk is present in concentrated form. A worker converting to fund retirement in a year like 2022 would have converted at a fraction of the prior year's value. Liquidity is not the constraint. Price at the moment of need is, and no depth of market removes it.
The Tension That Cannot Be Engineered Away
Notice what happened between Ownership and Human.
The form that passes Ownership at its maximum is the form that fails Human hardest. The second paper's governed-property resolution is the only route through: held through a custodian, inside a plan, with defaults, automation and institutional stewardship of the decision layer, the inaction path could in principle succeed.
But governed holding is holding whose custody has been converted from property to promise. And the second paper's standard bites hard here: an instrument that offers claim security only at the price of demanding operational control from the worker has purchased its ownership virtues with the currency the worker does not have.
This asset faces that standard more sharply than anything else in the inventory, because its ownership virtues are its principal case and its operational demands are its principal failure.
Custody design can position the holding between the poles. Multi-signature arrangements, qualified custodians, collaborative custody with recovery paths, fiduciary structures that steward the decision layer while leaving the claim layer nearer the worker. Each is a genuine contribution.
None dissolves the tension. Every mechanism that reduces the worker's operational burden reintroduces a party whose performance the worker depends on, and every mechanism that reduces that dependence returns some burden to the worker.
The trade can be positioned. It cannot be escaped. I record that as a finding rather than a problem awaiting an engineer.
It Is Not a Pension
This deserves saying without hedging, because the argument will be misread without it.
It does not provide mortality credits. It does not pool longevity risk. It does not insure against disability, the shock that ends the human capital funding the whole chain. It does not deliver guaranteed retirement income, or any income at all; it is a stock rather than a flow, and it becomes a flow only by being sold, which places it on the wrong side of Merton's insistence that the worker's true objective is income rather than wealth. It does not supply behavioral governance. It does not represent ownership of productive enterprise. It does not protect against healthcare costs. It offers no certainty of purchasing power at any specific future date, no survivor benefit or beneficiary continuation beyond what estate planning supplies, no spouse or dependent protection, and no indexation of an income stream, since it produces no income stream to index.
Every one of those is a function the second paper identified as necessary to a mature preservation architecture, and this asset performs none of them.
So it is not a preservation architecture, and it is not a pension. A worker holding it and nothing else would have an unassembled architecture in the most extreme form the second paper described. The proposition that it could replace pensions, or that a worker could retire on it, is not one I entertain and not one the framework permits.
Read across the six tests, the pattern is legible, and it is not the pattern of a solution. It is a failure pattern in the second paper's sense, and a distinctive one. The asset is strongest exactly where the second paper's inventory was weakest, on issuer independence and claim portability. It is weakest exactly where the inventory was strongest, on human usability, pooling and horizon reliability.
It is the mirror image of the collective architecture the second paper scored highest. And the mirror image of a strong incumbent is not a replacement for it.
At most, it is a complement whose function is defined by what the incumbents lack.
Next week, the last one: what that function actually is, what it is not, and the bounds the framework can put on it.