This essay is one of many published in Decentralized Thriving, an ebook covering the current state of decentralized governance.

On Regulating Decentralized Systems

With the advent of the internet and now blockchain (& its ilk), it’s tempting to imagine decentralized, peer-to-peer systems overthrowing all the old hierarchies. In the decentralized future, there will be no banks; there will be peer-to-peer currency protocols. There will be no companies or governments; instead, we’ll all be members of various DAOs (decentralized autonomous organizations). National borders will lose their meaning, and a more Pareto-efficient network will connect people all over the globe according to their individual abilities and needs. This vision of a decentralized future is in the air at “crypto” conferences and meetups, and though most would probably be hesitant to publicly endorse it, there is a sense that these new decentralized systems have limitless application.

Nothing has limitless application, of course, and in the case of these new decentralized systems, they have a limited ability to address negative externalities. In some cases, these systems are even predisposed to exacerbate their negative side effects. This is not a new problem. Over the past two centuries, private companies have often been accused of the same flaw, that is that their quest for profit has unwittingly caused some broad societal harm, like children injured doing hard labor or carcinogens in the water supply. Decentralized systems like Bitcoin have many of the same financial incentives as private corporations but are more resistant to traditional regulation, and so the question of how to limit the systems’ negative externalities is not trivial.

I’ll argue here why externalities in decentralized systems are a looming problem and suggest directions in the search for solutions.

Decentralized systems

The definition of “decentralized system” used here covers more than just Bitcoin, Ethereum, and the other distributed ledger technologies. A decentralized system is made up of any number of components, the system’s nodes, that make choices independently according to their own local information. Decentralized systems are a lot like games. Each node is a player trying to achieve their own ends, constrained by the game’s rules but no active authority. The total result of a decentralized system is the sum of the results of each node’s actions.

Systems and nodes. “Distributed” systems are considered a subset of decentralized ones in this paper, since both have the qualities ascribed to decentralized systems here.

Bitcoin fits this definition. Each “player” chooses whether or not to run Bitcoin software by weighing the costs and benefits to them personally. Evidently, the benefits have been compelling enough to convince many people to participate, and the cumulative result is the functioning Bitcoin ledger.

Bitcoin belongs to a special category of decentralized systems that also have decentrally-chosen rules. Not only does each node decide what to do within the constraints of the system, but the architecture of the system itself is decided in a decentralized way. There is no central committee that can declare the correct version of the Bitcoin node software. Instead, the “correct” version is whichever version the nodes decide to use.

Kids playing basketball at a playground is also a decentralized system with decentralized rules. Each kid makes their own choices when playing, but they also come to their own location-unique consensus about the rules of the game, which is why you’ll find some strange versions of basketball on playgrounds around the world. You’ll never find anything but the standard rules in an NBA game, however, because NBA basketball is a decentralized system with centralized rules. A commission sets the rules, and then the players make their own choices during each game (with the exception of coaching). A few more examples:

  • Cultural norms like saying “please” and “thank you” are decentralized systems with decentralized rules. No single body decides when the socially acceptable times to use these phrases are: the rules emerge from our collective behavior.
  • Twitter and Facebook are decentralized systems with centralized rules. You post what you want, but they set the character and content limits.

Controlling the rules of a decentralized system is a great way to control the system’s overall outcome: the NBA commission has had success changing the rules of NBA basketball to make it more and more entertaining for viewers. In a decentralized system with decentralized rules, neither the nodes nor the rules are easily controlled by any niche group. As we’ll see, that makes it difficult to fix chronic problems in such systems.

Decentralized systems with decentralized rules can have significant negative externalities

Bitcoin and Ethereum are two of the largest new systems with decentralized rules (market capitalizations measured in billions of U.S. dollars), and both have been accused of causing significant negative externalities: harmful side effects the systems’ rules don’t take into account.

In economics, externalities come from differences between a good’s social cost (e.g. its effects on the environment or on your neighbors) and its private cost (its price).

Some negative side effects touch only the systems’ own nodes. One such effect is the re-centralization of the systems’ networks. Both Bitcoin and Ethereum (at the time of this writing) use Proof of Work consensus algorithms that reward nodes running mining software on the fastest, most expensive hardware. People running the most powerful mining nodes earn more money, which they can reinvest to increase their advantage ad infinitum, progressively eroding the resilience a decentralized system provides.

Another participant-exclusive externality is the level of risk imposed on nodes. Cryptocurrencies are prone to wild price behavior, and that has been especially true over the last two years.¹ This volatility means nodes and token holders must now accept high risk: major asset appreciation and depreciation are both uncomfortably likely.

Other externalities extend to more than just the systems’ nodes. It’s possible that Bitcoin is exacerbating global wealth disparities, for one. About 41% of all wealth of any kind was owned by 0.7% of people in 2013.² That’s a high concentration, but it’s only about half as concentrated as Bitcoin, where 87% of all bitcoins are owned by 0.66% of addresses.³ If such wealth statistics represent a real problem, then Bitcoin may worsen that problem severely as its adoption progresses.

A second externality affecting the wider world is energy use, an oft-criticized facet of Bitcoin and Ethereum. The Bitcoin system apparently consumes about as much electricity as Portugal, and it’s not at all clear if the system’s benefits are worth that cost.⁴ Given current energy consumption, it’s ominous that Bitcoin may still be at the beginning of its growth curve.

Some externalities imply a need for outside regulation

The nodes of a system with decentralized rules can choose to change their system to reduce externalities, but they’re only likely to do so for externalities with significantly negative impacts on the nodes themselves. Theoretically, decentralized systems, like other systems, can be designed to mitigate any externality (to the point of simply shutting down the system if the externality is negative enough and cannot be otherwise eliminated). But “can” doesn’t imply “will” in theory nor in practice.

For Bitcoin and Ethereum, which hypothetical externalities pose the greatest threats to their system’s nodes? In theory and in historical precedent, externalities affecting much broader groups pose only a diluted threat to nodes.

  • Global wealth disparity — As long as lopsided world wealth distribution or even wealth disparities among Bitcoin and Ethereum holders don’t undermine the systems’ security or reliability, nodes have little reason to mitigate it. The systems are functioning fine today despite high wealth concentrations.
  • Global environmental impact — There are no internal penalties in Bitcoin or Ethereum for increased electricity usage. In fact, miners are rewarded for using more electricity in many cases, so if anything, the systems’ energy costs and environmental footprints may only get larger over time.

Externality effects concentrated on the system’s participants are more likely to seriously threaten them and to spur action.

  • High risk for investing nodes — Return on investment is one of the main motivations for nodes participating in these systems, and risk is a deterrent to investors. If risk goes too high and nodes start leaving, remaining Bitcoin and Ethereum nodes will have strong reason to adjust system rules.
  • Re-centralization of the system’s infrastructure — The independence of Bitcoin and Ethereum from any central governing bodies is a major selling point for most of the cryptocurrency community, so too much re-centralization of power is a reasonable motivation for nodes to change the system.

The track records of companies and governments support this pattern. While people have always tried to take care of their own backyards, the development of environmental policies by companies, states, and global alliances has been halting at best.⁵ Likewise, private corporations as a whole are addressing employee health risks voluntarily ⁶ but overall community health risks (like water pollution) only under duress from regulators. For a specific case, think of Walmart raising minimum wage for its workers but ignoring the way opening a new Walmart store can starve a town’s other businesses.

Externalities that are ignored are likely to be exacerbated. Bitcoin, Ethereum, and most other newly prominent cryptocurrencies have the same infinite growth incentive as public corporations: no matter how large they are, stakeholders continue to gain from further growth. Bitcoin and Ethereum holders will always gain from higher demand for their networks’ tokens, and so they are likely to continue pushing adoption as long as negative externalities affecting them don’t pile up too high. As these systems continue to grow, any side effects they don’t eliminate will grow as well.

If decentralized systems

  1. don’t adjust for all externalities,
  2. worsen negative side effects they don’t fix, and
  3. may cause alarming consequences via these unaddressed effects,

some form of external regulation will be required. Because we are speaking of decentralized systems, however, regulation cannot use its standard playbook.

Regulatory bodies for decentralized systems: lawmakers, influencers, or DAOs?

As decentralized systems with decentralized rules, Bitcoin, Ethereum, et al are hard to regulate like culture is hard to regulate. They are swarms, acting collectively via stigmergy, not hierarchy. It isn’t obvious which outside mechanisms can best influence these systems, but three directions seem reasonable:

  • Legislation (proven but costly)
  • Advocacy (flexible but unreliable)
  • “Umbrella” DAOs (powerful but theoretical)

Though it’s no one’s favorite option, governments can regulate decentralized systems if those systems depend on centralized infrastructure or the systems’ nodes can be surveilled. The core of Ethereum may be censorship-resistant and peer-to-peer, but its community relies on identifiable figureheads and centrally-hosted exchanges and social networks: these things can be targeted by governments as an indirect-but-potent way to regulate Ethereum. Further, governments are capable of surveilling and penalizing individual internet users, as when the RIAA (Recording Industry Association of America) was permitted to track and sue thousands of individuals who had pirated mp3s. This approach has pitfalls: the RIAA had to switch tactics after 35,000 cases over 5 years tanked their public image and failed to sufficiently discourage piracy. Evidently, intense government regulation has its own externalities.

Regulation through advocacy might be a more agreeable option. Culture seems to move when loud, persuasive voices push it, and the same might be true of other decentralized systems. We’ve seen the standard language on race and sexuality shift as individuals and organizations have advocated for new ways to understand those concepts. As in those examples, the best way to reduce Bitcoin’s wealth concentration (if it is indeed a problem) might be to convince Bitcoin nodes to change their values. The catch is that advocacy is an unreliable change-maker. Persuasiveness cannot be manufactured, and even if a message does go viral, tracking its effects is guesswork. Even still, the plausibility of advocacy as a regulatory method is a reason to applaud (and join!) those currently campaigning to shape the values of decentralized systems.

A historical example of changing the rules of a decentralized system through advocacy. In this case, changing some norms of Hong Kong culture through public protests (the Umbrella Revolution).

A third regulatory option is “umbrella” DAOs. Decentralized Autonomous Organizations are a type of decentralized system that is able to form system-wide consensuses on arbitrary topics (usually through voting). Umbrella DAOs with wide membership and jurisdiction over smaller, “child” decentralized systems might be able to directly regulate them. Because DAOs are also decentralized, granting one direct control over the software used by another decentralized system’s nodes does not bring the risks of centralization. An umbrella DAO’s members would need to include both child system members and the people affected by whatever externalities the child systems are ignoring, so the the umbrella DAO would presumably have reason to respond to those externalities. While individual systems might never fix their environmental externalities, an umbrella DAO including more people suffering from climate issues might act. The idea seems promising, but the first DAOs have only just come into existence, and their basic soundness is still unproven. Creating these umbrella DAOs would also be a major challenge: why would systems agree to be under their jurisdiction? What would fair representation in umbrella DAOs look like?⁷

These regulatory tactics may prove to be important. The potential of blockchain and other new technologies to unlock new forms of coordination is undoubtedly exciting, but these systems are also making large-scale collaboration more autonomous than ever. Autonomy is double-edged, and as these new systems’ influence spreads through society, we should be ready to compensate for their unintended negative effects. Legislation, advocacy, and DAOs may all be useful in doing that. ☂️

This essay is one of many published in Decentralized Thriving, an ebook covering the current state of decentralized governance.

¹ Since November 2017, Bitcoin’s market price has zigzagged from ~$6,000 to ~$15k, ~$8k, ~$11k, ~$7k, ~$9k, ~$6k, and ~$8k. In September 2018, it seemed to settle around ~$6.5k, until November brought a further plunge to ~$3.5k. Ethereum and its token, despite being a separate system, have followed a nearly identical trajectory.

² Estimated by Credit Suisse.

³ As of January 30, 2019.

Portugal and Bitcoin are both estimated at around 47 tWh / yr. Ethereum is at about 7 tWh / yr. These estimates may be rather inaccurate, but my argument is also crude: these systems use a lot of energy relative to the rest of their impact.

This article by Somini Sengupta reflects some common views on the topic.

⁶ “Voluntarily” as in changing because of competition for employees or because of unions, rather than being forced by authorities.

⁷ Many decentralized systems (and systems that are not currently decentralized) may use DAO structures in the future, and from a regulatory perspective, this would be useful. Because DAOs help decentralized systems arrive at reliable consensuses, system using DAOs will make more defined, predictable decisions than systems not using DAOs. Any tactic to regulate these systems should be easier to implement on a system with more consistent behavior.