Knut Karnapp posed this very interesting question over on Twitter. His answer:
To me you own a part of the Bitcoin UTXO set uniquely assigned to you, and only you — by virtue of the corresponding private key. With this comes great responsibility. If you lose your private key, you lose your bitcoins. If your private key gets stolen civil law may dictate that the key itself and the UTXOs accessed by it are still “yours”. As far as the Bitcoin network is concerned though the private key grants power of disposition to whomever is in possession of said key.
That’s a solid answer from a de facto point of view, where continuing knowledge of the private key basically == what most people commonly refer to as control, or ownership. From a workaday transactional standpoint I basically agree with it wholeheartedly. De jure, on the other hand…
“Ownership” is more than mere control; it is a legal concept and law is a local phenomenon. Accordingly, when you ask yourself whether and how something is owned, it’s generally a safe assumption to begin, in the first instance, by looking at the governing law of the asset and asking what that governing law says.
With certain things, like securities, the governing law of the issuing jurisdiction/entity and the governing law of the instrument (if different) are likely to be dispositive. International bearer securities, e.g., utilize well-worn issuance frameworks like the New Global Note structure, which divides up legal and beneficial title in the underlying security by contract in a manner that is highly certain and leaves little room for ambiguity. With real property (an apartment, a house, some land) you usually look to the law of the situs as the starting point for that inquiry. Generally speaking it’s the same story for chattels, save where ownership of those chattels is represented by a certificate of title or the like.
The problem with Bitcoin, of course, is twofold.
First, Bitcoin does not avail itself of existing categories of property, like chattels or instruments; indeed, it defies them in many respects. As a consequence, any contractual or systemic understanding of the thing – to the extent one exists at all – is going to be implied, and seeing as courts haven’t really grappled with foundational questions about what Bitcoin is, we don’t know what that implication will look like. The best we can do for now is guess what the boundaries of that implication, once set down in writing, will be. We will call this the Classification Problem.
Second, a bitcoin does not really have a physical location, and is a fundamentally global good – it exists on every computer which runs a full node, and is arguably issued everywhere and nowhere at the same time. But the Classification Problem will be determined by reference to local, not global, rules. We will call this the Forum Problem.
The “Forum Problem” is a simple one; Bitcoin has no identifiable origins, no clear home, so each different country/jurisdiction in which litigation over Bitcoin is brought (in the case of the U.S., the states and the various federal jurisdictions) will feel entitled to apply its own rules to the asset. For the majority of commercial arrangements, harmonization can probably be achieved by choice-of-law clauses among the counterparties to the transaction.
The “Classification Problem” is where things get more interesting. Here we ask what rules each jurisdiction would apply if some litigation arose which involved fundamental questions about the nature of ownership as it pertains to Bitcoin the asset. Usually, those fundamental questions are not in dispute in the kind of workaday litigation that comes before the courts. Courts take judicial notice of who owns what bitcoins based on the facts of the case; Alice sold some bitcoins to Bob, there are no competing claims to the bitcoins and the question is whether one of the parties reneged on the high-level commercial terms of the deal.
What hasn’t happened yet, and what invariably will happen as more and more cases hit the courts, is that someone will ask the question, “what property classification do we apply to Bitcoin – WTF is it that Bob actually owns?” This is because, at its core, a bitcoin is really, in its purest essence, only a solution to a randomly-generated math problem. Because the problem is very hard, the combination of a UTXO plus the ability for a recipient to spend it, armed with the knowledge of the relevant private key, is treated by most of us today as property. That “property” creates a write permission on a massively distributed cryptographic ledger which nobody controls, although control of that write permission can be transferred to other users of that database by spending the associated coins to those other users.
Because the secret embodied by a private key one does not know is very difficult to obtain – and impossible to obtain on a commercial timescale with existing technology – people call Bitcoin a “digital bearer asset.” Bitcoin is most assuredly not a bearer asset or chattel, though. Nor is it a documentary intangible, as it is not a contract and is silent, apart perhaps from the provisions of the MIT Licence, as to what a court should do when presented with one (more on that below). Unlike physical goods which can only exist in one place at one time, it is conceivable that with a powerful enough computer, the solution could be found entirely honestly by a third party simply doing some math and stumbling upon the answer at random, or by asking the right questions and exploiting some as-yet-undiscovered weakness in the implementation.
Bitcoin might, therefore, be better described as a digital I-know-something-you-don’t-yet-know asset. “Yet,” because the information is not secret (in the same way as a trade secret, e.g.) or impossible to ascertain; it’s out there waiting to be ascertained by someone clever enough, or a computer powerful enough, to figure it out. The term “cryptoasset” that is cavalierly thrown around by your garden-variety ICO bro inadvertently turns out to be an accurate description for this new class of ownership. Lawyers wishing to confer dignity on the phrase might call it “crypto-property” instead.
If we really wanted to make it our own and de-emphasize the “there’s a lot of cryptography in this thing” aspect of Bitcoin, which is not legally relevant, in favor of an laser-focused emphasis on exclusive knowledge of the secret key as being dispositive for control and highly relevant for ownership, I might suggest the radical step of changing the spelling of the prefix to “krypto,” per the original Greek κρυπτῷ, so we’re left with “krypto-property.”
Contrasting approaches between England and the U.S.
Who owns the solution to that really hard to solve, but solvable nonetheless, math problem? I ask this question, which seems like an obvious or even pedantic one, only because I am fairly certain that the world’s two largest common-law jurisdictions – England and Wales, and the United States – would reach different conclusions.
Now, of course Bitcoin is treated as various things by various agencies of the state – most significantly, as “property,” by both the IRS (American taxman) and HMRC (English taxman). But that doesn’t answer the question of what kind of property the stuff actually is.
In England, for example, longstanding precedent has held that “the right to confidential information is not intangible property;” see Oxford v. Moss, (1979) 68 Cr App Rep 183 (a student cheating on a test by reading the answer sheet in advance could not be convicted of theft, as the answer to the test – as pure information – was not intangible property and therefore incapable of being stolen). This principle nukes the notion that a private key is worth more than the paper that you [really should not] have written it on.
At the same time, English law may have an equity, which looks a lot like a property interest in confidential information that has been misappropriated, that gives a party wronged (i.e., for our purposes, the person from whom knowledge of a private key was wrongfully obtained) a right to restitution. Anyone looking for the detailed treatment should read the section “Information as Property” at page 1 in Palmer & McKendrick’s Interests in Goods (1998).
I wrote a fairly lengthy analysis on the English law in this area back in the day, which, annoyingly, I have since lost. TL;DR, if an attacker fraudulently obtains a private key, English law provides a a remedy, but if an attacker should stumble upon a key by accident or by brute force, it probably doesn’t. This is unsatisfactory but it’s what we’ve got.
Contrast this with the U.S. position, where the courts have found that property rights can subsist in pure information such as unpublished or recently-published news (INS v. Associated Press, 248 U.S. 215 (1918)) or straightforwardly analogize doctrines such as relativity of title as a hack/workaround (e.g. Popov v. Hayashi, WL 31833731 Ca. Sup. Ct. 2002). Incidentally, a relativity-of-title-theory approach would also solve, for most practical purposes, what the inimitable Izabella Kaminska described as “Bitcoin’s Lien Problem” in 2015; it strikes me that that theory of ownership should be fairly good fit with UTXO-based systems where one can trace title to a given coin perfectly, give notice of theft or fraud efficiently, and prove current “possession” with a high degree of precision. Crucially, it might prevent an attacker – even an accidental one – from getting superior title to the Bitcoin he obtains, as long as the courts or the legislatures decide that’s how they want to crack that nut.
Equally, and another idea I have noodled on, is that Bitcoin’s code is really the first “smart contract” in that the code embodies a binding contractual understanding among the users. However, the fact that the code can be forked by consensus of the users to say anything at any time suggests to me that a court would likely conclude that there was not a clear intention to create a contract by running the code and so might refuse to enforce a particular mode of operation on the users of the network (see e.g. Jones v. Padavatton,  EWCA Civ 4). Incidentally this absence of a contractual understanding/effective ousting of the jurisdiction of the court is why Bitcoin cannot and should not be described as a chose in action.
Wrapping up, the reason that the matter of Bitcoin’s ultimate classification as property hasn’t come up yet is because, in common practice, ownership disputes are resolved at a higher conceptual level than inquiring about the “nature of a bitcoin itself” – when I deposit coins at an exchange, e.g., it ought to be pretty clear from the exchange’s TOS that if I have a balance on the exchange, I can ask the exchange to spend an amount equal to that balance back to me on request and, if they fail to do so, I can ask a court to force the exchange to render specific performance or pay damages. A dispute of that kind, of which there have been many, doesn’t ask at what point title transferred and what the fundamental nature of that title is, because it doesn’t have to. It looks instead at the contractual obligations between the counterparties and whether those obligations were satisfactorily performed.
One could write chapter and verse comparing these two jurisdictions and their treatment of Bitcoin as an asset. That said, it’s a Friday night and I have places to be, so for now it will have to suffice to say only that the question has no answer and at some point, probably sooner rather than later, there is going to be a case that explores these fundamental issues (I am frankly shocked that Oxford v. Moss hasn’t been raised yet in any of the UK-based Bitcoin fraud prosecutions).
I look forward to reading those decisions.
Too good not to share.