The Earl of Ethereum’s CaseWho keeps a blockchain’s conscience?
Earlier this year, a company called Slock.it wrote, released, and ran a computer program called The DAO. The DAO was designed to function like a seed-stage venture capital firm, soliciting capital, taking investment decisions, remitting returns to investors, and administering back-office processes. Receive capital it did.
The DAO ran not on one computer, but on many, using a system called Ethereum. Ethereum permits many computers belonging to many people to perform and verify every software instruction followed by a computer program, like The DAO, calculating for themselves the effects on the accounts of each participant. Each account is protected from basic frauds, like double spending.
As it turned out, The DAO’s instructions created a loophole. The loophole was discovered. Ethereum users watched an anonymous peer carry off capital from The DAO’s digital accounts, all in strict compliance with its code, independently verified by a multitude of willing, diligent, and indifferent machines, many belonging to those defrauded.
Troubled by the outcome, a large group of Ethereum users agreed to introduce an otherwise invalid transaction to reverse the effect of invoking The DAO’s vulnerable code. This initiative stood each Ethereum user at a metaphorical fork in the road. Either they could join those following The DAO as it played out, by its terms, or they could join the larger group making a specific exception, overriding Ethereum’s rules to avoid The DAO’s outcome.
Every part of this process has been noisy, controversial, and self-consciously philosophical. The energetic Ethereum community is not united by wonkish obsession with the back-offices of venture capital firms or the arcana of corporate governance and securities law. Rather, they rally behind the promise—prospect—of computer programs that run reliably and verifiably, unbeholden to and unassailable by human fallibility and interested-party control of computing infrastructure. It’s a vision of governance of laws and not of men, with the laws all in code and the men, if men there must be, tightly confined to narrow corridors of well-understood economic incentives, like cattle in a chute.
The more politically and philosophically oriented elements of the community aim to break with troublesome aspects of the prevailing order, from the boom-bust-bailout of the financial cycle to pervasive surveillance, regulatory interventionism, and the prohibitive cost of justice. Casualties be damned, that Ethereum should operate like so many meaty, judgment-dependent institutions of yore irks them mightily. As does the stampede of blockchain start-ups, often VC-funded, scurrying to apply breakthrough technology to more banal problems of banks and accounting firms handling so much money they can’t reliably count it all.
The quotes are the best parts of this post. If you’re going to skip something, skip what I wrote and read the quotes:
First, As a Right in Law cannot die, no more can Equity in Chancery die, and therefore [no one should depart from Chancery without a remedy] … for Conscience and Equity is always ready to render to every one their Due …
The Cause why there is a Chancery is, for that Mens Actions are so diverse and infinite, That it is impossible to make any general Law which may aptly meet with every particular Act, and not fail in some Circumstances.
The Office of The Chancellor is to correct Mens Consciences for Frauds, Breach of Trusts, Wrongs and Oppressions of what Nature soever as they be, and to soften and mollify the Extremity of the Law, which is called [highest right].
And for the Judgment, &c, Law and Equity are distinct, both in their Courts, their Judges, and the Rules of Justice; and yet they both aim at one and the same End, which is, to do Right; as Justice and Mercy differ in their Effects and Operations, yet both join in the Manifestation of God’s Glory.
The Chancellor sits in Chancery according to an absolute and uncontrolable Power, and is to judge according to that which is alleged and proved; but the Judges of the Common Law are to judge according to a strict and ordinary (or limited) Power.
— The Earl of Oxford’s Case (1615) 1 Ch. 1, 21 E. R. 485.
Early English common law was infamously complex, fiendishly exact, and coldly unfeeling. At least in principle, it was not up to judge, jury, or attorney to account for the evident injustice of a properly derived legal result. Like a great machine turning over, the courts were meant to receive the law, tracing its authority back to the king’s, and apply it to cases as a mechanized brewery applies bottle caps to new brewskies. The process was not entirely mechanical—the law came down and moved around in unstructured form, a mad babel of Law Latin, Law French, and bastard English, often unwritten—but the view of its proper function often was mechanical. All in keeping with the elemental, if exception-ridden, notion of monarch as font of all property rights.
Those run over roughshod by the common law could appeal, as one might appeal today for review or a pardon, for the direct intercession of the monarch or his confidants. The common law being what it was, such requests mounted, and the duty of reviewing and responding to them was similarly delegated to a specialized servant of the crown, the Lord Chancellor. The Lord Chancellor was meant to embody the King’s softer, conscientious, divine-representative side, with corresponding authority over souls, as distinct from property. The Lord Chancellor got busy and so begat his own court, the Court of Chancery.
The Earl of Oxford’s Case, decided in 1615, established the primacy of the Chancery Court’s “equitable” orders, grounded in often fuzzy and naturalistic principles, over those of the courts of law. In other words, the Court of Chancery could issue orders, called injunctions, against performance of legal orders and further suits in the courts of law. Equity became the flexible yin to law’s rigid yang. Hard law made a strong case for flexible equity, which expressed its gratitude by making equitable decisions seem squishy, arbitrary, and corruption-prone in comparison to legal ones.
The distinction between law and equity echoes in the English legal system and its derivatives, including America’s legal systems, of today. Remedies—actions you can ask a court to take on your behalf—are divided into legal remedies, like the payment of money damages, and equitable remedies, like orders to perform your part of a deal, transfer an interest in property, or refrain from specific conduct, all backed up by penalties for contempt of court. Old principles of equity, oft quipped as pithy “maxims”, even in modern court decisions, still apply to some equitable remedies. He who comes into equity must come with clean hands. Equity will not allow a statute to be used as a cloak for fraud.
Both English and American legal systems have nominally collapsed law and equity into a single system, with courts hearing arguments based in both legal and equitable principles, and handing out remedies of historically legal and equitable pedigrees. A few states and federal subject areas maintain specific equity courts, some very much on theme, like United States bankruptcy courts, and others for more dubious and idiosyncratic reasons, like Delaware’s Court of Chancery, which takes first pass at vitally important corporation law issues.
In the main, the trend is toward single courts that do the business of justice without regard for the old dichotomy. This is often far more efficient: It’s fairly clear where you ought to bring your lawsuit, and easier to juggle one set of rules, procedures, and schedules once you get there. On the other hand, it’s sometimes hard to tell whether a court is refining and applying rules for strict application or making room for itself to avoid an unjust result. This has been especially clear in contract law of late, with many courts straining to invalidate “contracts of adhesion”, like dense online terms of service, and specific provisions, like commitments to arbitration that deny consumers access to the public courts. Some courts have even seemed to play games with ambiguity and vagueness, seizing on whatever weakness they can manufacture to avoid doing The Wrong Thing.
Legal practice has a habit of deep-freezing vocabulary and ideas well past their expiration dates. But as often as I’m a critic of current legal practice and style, I think there’s something more to the particular holdover of law-versus-equity. There is something intuitive, archetypical in the idea of salvation from a cold and unflinching rule. I’m not sure one mind responsible for both can do as well as two minds, one framed to do each.
How many Russian serfs must have cried out “If only the tsar knew!” in exasperation before 1905. How many of us clamor now for a Snowden pardon in the twilight of another presidential term. How many VC funds and other companies settle internal disputes by written compromise or amending their governing documents. A long time ago, we discovered that rules and a predictable way to apply them make a powerful antidote to individual bias and corruption. But we’ve yet to build any system on that first principle that doesn’t chew through good people for bad reasons on a regular, if tolerably infrequent, basis. We’ve always needed an escape valve, and omitted one at our moral peril.
But what can be more splendid than a state governed by worth, where the man who gives orders to others is not the servant of greed, where the leader himself has embraced all the values which he preaches and recommends to his citizens, where he imposes no laws on the people which he does not obey himself, but rather presents his own life to his fellows as a code of conduct?
If one man alone could meet all these requirements there would be no need for more than one.
— Cicero, The Republic, Book 1, 52. (Niall Rudd, trans.)
There is a kind of technology in recognizing, accepting, and coping with the malice and incompetence of human agents we have to trust. To their credit, the Ethereum community has never artificially limited its problem to its software, or even the network running its software. Vitalik Buterin’s writing has considered incentives to fork and the likelihood of programming errors from very early on. He was not alone. Optimism has definitely been a constant, but modulated by a commendable respect for the limits of current knowledge and the dimensions of the problems they’ve opted not to skirt.
To this lawyer’s eye, the most interesting thread in that thinking has always been “proof-of-work” versus “proof-of-stake” systems.
The best-known programs of Ethereum’s type, including Bitcoin, enforce rules requiring those who participate in deciding which valid decisions count and when to solve difficult computer puzzles as a kind of raffle ticket for temporary executive office. These are called “proof-of-work” rules.
The puzzles require lots of computer power, which costs lots of money. The more computers participate in the network, the more pricey computing power it takes to overwhelm the odds of all the other participants buying raffle tickets. Proof-of-work rules aim for the same effect—no small group of participants can take over collective decision making—without relying directly on expensive, often wasteful real-world challenges, like who can amass the most specialized computers or achieve the largest economy of scale in using them. Facts the network wouldn’t otherwise give a hoot about.
Proof-of-stake systems, by contrast, do not limit bad behavior by tying participation to a limited real-world resource. Instead, proof-of-stake systems levy procedural penalties against participants proffering transactions the network refuses to honor, in the form of reduced credits or privileges tracked within the network itself. Rather than requiring participants to spend real-world assets, like computing power, to buy executive office raffle tickets, proof-of-stake systems hand out raffle tickets, votes, or other powers to all, and punish their ill use after the fact, by a reduction in further right to participate. If those involved value ongoing participation on equal footing, therein lies their incentive to play straight. The effect is much like party discipline or adherence to rules of order at a meeting.
Proof-of-stake systems have proven plenty capable in many specific applications, especially within organizations that hold other levers of control over diverse mechanisms and incentives, and wherever the risk of collusion among double-spending fraudsters is low, even if the risk of malfunctioning software is real. In the lingo, malfunctions and attempted fraud are both “Byzantine faults”—invalid messages that computers can’t immediately recognize as bunk. A copy of the Ethereum program running haywire on a machine frazzled by hardware failure is a Byzantine fault. So is a program masquerading as Bitcoin, but designed to convince others to confirm transactions in an order designed to leave payees empty-handed. So, in a manner of speaking, are messages from Ethereum users on different tines of the Ethereum DAO “fork”.
However, to my knowledge, only proof-of-work rules have succeeded at large scale in systems, like the Bitcoin and Ethereum networks, without overarching checks and balances—derisively, “centralized authority”. Such use cases, in the aggregate, probably make up the vast majority of useful applications of the underlying Bitcoin-esque technology. But for those of a more political bent, those use cases are precisely the uninteresting, even repugnant ones.
Lack of central authority without the wastefulness of proof-of-work rules is the techno-political dream, in that it would make the system wholly self-contained, a kind of sealed Bottle Garden of Economic Eden. Such a system might be used to make agreements or administer organizations that interact with the real world, say by performing services or dealing in goods. But their software-automated bits could be verified and audited by the network independently, guaranteed by the shared interest of others utilizing the network in accurate record keeping.
In theory, that shared interest might be enough. If it works there’s no need to interlock the system with other possible sources of corrupting leverage—financial, technological, political. That being the case, the decentralized system clearly obviates all need for existing centralized, evidently corruptible analogs. It becomes both “disinterested” in the sense of being incapable of having a selfish interest in the outcomes it administers, and “incorruptible” in the sense that no outside leverage can be applied to induce such an interest.
The question Whether one generation of men has a right to bind another, seems never to have been started either on this or our side of the water. Yet it is a question of such consequences as not only to merit decision, but place also, among the fundamental principles of every government. The course of reflection in which we are immersed here on the elementary principles of society has presented this question to my mind; and that no such obligation can be transmitted I think very capable of proof. I set out on this ground which I suppose to be self evident, “that the earth belongs in usufruct to the living;” that the dead have neither powers nor rights over it.
— Thomas Jefferson to James Madison (1789)
Buterin pointed out years ago that questions about proof-of-work and proof-of-stake systems boil down to broader questions about how much we value “decentralization”—freedom from overarching, “centralized” checkers and balancers that fail in manifold, often obscure, ways. Proof of work is expensive, but that expense may be worthwhile. Hope for a viable alternative would spare us that valuation riddle. What we can automate safely at low cost and negligible risk, we will be tempted to automate. We’d rather cut the Gordian knot than untie it or live with it.
The prospect raises two important questions.
If we find what appears to be a suitable sword, can we trust that it will do the job? How can we verify that the job is done? Every durable system of governance I know of—legal, political, governmental—has installed within itself an escape valve, a means of correction when established rules, predictably, fail to function adequately in unforeseen circumstances. Even such safeguards routinely fail. It’s no easier to architect the perfect mechanism of reform than it is to architect the perfect substantive rule you were after in the first place. This is a humility every practicing deals lawyer learns, hard or easy. The same can’t be said for every programmer.
Equity’s primacy over law. Procedures for termination and amendment. The institution of the “dictator”—originally limited to a specific dictate and extraordinary circumstances. These mechanisms help to avoid the costs and uncertainties of rebuilding anything from an entirely clean slate, though that is always a possibility. As far as blockchains go, the answer is probably that failure in these early days is livable enough to warrant free experimentation. If a system fails, we’ll survive, and we can learn from it. If the line of inquiry trails off, it won’t have been the first honest fishing expedition. Blockchain as a whole might be a fishing expedition, at least as far as the cryptoanarchists are concerned.
The harder question is whether, having convinced ourselves that we’ve found a way to solve these governance problems in utter independence form the world around us, we should use it.
The freedom-enabling, automobile-like elan of the computer is a late-model marketing invention. The computer more generally made its debut as a tool of systematic dehumanization and oppression, the conscience-free mechanical lackey of indifferent captains of industry and McNamara-ite technocrats grinding the Bartlebys of the world into mincemeat. Levy’s Hackers describes anti-war protests on the floor of MIT’s AI Lab, then home to a Department of Defense-funded projects and hackers bewildered that their peers should associate computers with the war. These days, industry and government’s taste in sprawling systems accessed by low-margin consumers via relatively dumb devices has produced all kinds of security and privacy issues that are beginning to unwind the glossy-full-page-ad view of the computer once again. Computers are seeming less cool all the time.
There is arguably no manipulation in today’s politics so terrifyingly opaque and invasive, for those outside the cabal, as the prospect of cryptoanarchists “engineering” their incentives, binding them to structures obscure to anyone else, and only partially understood by any individual cryptoanarchist. That’s the kind of unchecked prerogative we expect, but don’t approve of, from ad-men, politicians, and financiers, who have been bad enough without the advantages of technical acuity. It is frightfully easy to see the benefits and liberties unlocked by one’s work, maddeningly difficult even to remember to look for the limitations and inequities.
I support these experiments, and I support the experiment that was The DAO. Finding a way to make these systems self-sufficient, by fully decentralized proof-of-stake rules or otherwise, would be a startling achievement. But I worry that, if progress here catches hold of the really fundamental governance problems implicated, it will roll right over so much of what we know about governance more generally, some of which is truly ancient knowledge. I worry that it will collide with strong institutions more by frenetic, Brownian motion than reasoned engineering, even if plenty of cryptoanarchists, young and old, will be ready to say that this, finally, is the blockchain they always wanted and expected.
Your thoughts and feedback are always welcome by e-mail.