Eli Dourado

The case for Ethereum maximalism

Preface: Don't @ me. I have opinions. You have opinions. Our opinions might not be the same. That’s ok. There is no need to tell me how stupid or evil I am. If you feel a visceral reaction to this blog post, you might explore the possibility that you hold a tribalist attachment to your cryptocurrency views and wonder whether this is really healthy or truth-tracking.

So far, three cryptocurrencies—Bitcoin, Ripple, and Ethereum—have hit a market cap of $100B. Notably, they all did so within three months of each other. These cryptocurrencies first hit market caps of $10B in 2013, 2017, and 2017, respectively. Bitcoin reached the $10B milestone almost four years before any other cryptocurrency, but the $100B milestone only three months before its competitors. At least by this one admittedly imperfect metric, Bitcoin is losing its lead.

Bitcoin losing its lead makes sense to me. What doesn’t make sense is that Bitcoin still has a lead at all and that it still has so many proponents. In terms of actual utility, Bitcoin is inferior in almost every way to several other cryptocurrencies, most dramatically Ethereum. I find myself inexorably drawn to a position not publicly articulated or even necessarily held by Ethereum’s founders and most active developers—Ethereum maximalism. My cryptocurrency portfolio is invested accordingly, and I don’t need to justify my beliefs, but I am writing out my rationale for the benefit of friends and followers who are interested. Take it for what it’s worth.

The economics of cryptocurrency competition

My mental model of cryptocurrency competition is still the one Jerry Brito and I developed for our 2014 article on cryptocurrency in the New Palgrave Dictionary of Economics (here’s an ungated version, related blog post). One question this model addresses is whether you can simply clone a cryptocurrency and expect it to have much value. Our answer was generally not, because the cloned cryptocurrency will have the same technical feature set as the original currency but with inferior network effects and likely inferior governance.

More generally, you can think of each cryptocurrency as being characterized by a multi-dimensional set of attributes: security, transaction cost, network size, governance quality, robustness of scripting languages, and several others. My mental model is that when one coin is better than another on some attributes and no worse in any other, then it dominates the other coin. A dominated coin can have no value in equilibrium. The market is roughly winner-take-all between dominating and dominated coins (roughly because equilibrium isn’t achieved instantaneously).

This doesn’t necessarily imply the market will be winner-take-all generally. There may be coins that neither dominate nor are dominated by certain other coins. This can happen if cryptocurrency characteristics reflect tradeoffs between the set of attributes. It can also occur if certain characteristics are not unambiguously good. For example, more of one characteristic may be considered better for one application and less of that characteristic may be considered better for another.

These complications imply that single-chain maximalism is not a necessary outcome. But winner-take-all-ness between dominating and dominated coins used for the same application is a sure thing unless you want to bite the bullet and say that the value of all coins should fall to the marginal cost of creating them, i.e., zero.

Governance: a special feature

One especially important characteristic of a cryptocurrency are its governance institutions. How do cryptocurrencies decide to add new features? How do they decide to change network parameters in response to changes in computing technology? How do they address bugs that are discovered or crises that occur?

Hilariously, here is what Jerry and I wrote back in 2014 about Bitcoin in our New Palgrave piece:

Bitcoin currently has high quality governance institutions. The core developers are competent and conservative, and the mining and user communities are serious about making the currency work. An exact Bitcoin clone is likely to have a difficult time competing with Bitcoin unless it can promise similarly high-quality governance.

Since 2014, a lot has changed. Bitcoin has been unable to seriously address its on-chain scaling problems. Its community has alienated, marginalized, and purged dissenting voices, notably Mike Hearn, Gavin Andresen, and Jeff Garzik. Its core development team has been captured by an ideological faction committed to only off-chain scaling in the name of decentralization. This faction has undermined consensus scaling agreements and trashed the reputation of anyone who points out any of the above. As early as September 2015, I was concerned about Bitcoin governance quality, but still—mea culpa for that 2014 paragraph. I really got it wrong. (n.b., I have not consulted with my coauthor Jerry Brito on this post, and my views should not be attributed to him.)

Governance institutions are especially important for cryptocurrencies because they can’t be simply copied. You can perhaps copy the institutional structure, and you can copy the outcomes and decisions, but when a crisis occurs, you want the A team to handle it as calmly, reasonably, and professionally as possible. Source code and technical parameters can be copied. Adoption and network effects can be replicated over time. Good governance—like good culture at a company—is a challenge to develop, and once you lose it, it’s hard to get it back.

Ethereum vs. Bitcoin: Where do we stand?

Does Ethereum dominate Bitcoin? Not yet, but it seems likely to me that it soon will, at least in most important respects. Let’s look at some indicators.

Market cap

Bitcoin still leads in market cap, which I take to be a good indicator of adoption and network effects, by a factor of more than two.

Number of on-chain transactions

Ethereum currently processes more than 3x the number of on-chain transactions than Bitcoin does. This metric might be considered a proxy for the amount of real-world use.

Trading volume

Bitcoin still has about 3x the daily on-exchange fiat-currency trading volume that ether does. It seems fair to say that Bitcoin is still more liquid with respect to fiat currency than ether is.

Mining reward

Mining reward is a good indicator of how secure the core function of a blockchain is, as it correlates with the cost of conducting a 51% attack on the network. Mining rewards over a 24-hour period are about equal between Ethereum and Bitcoin, indicating that at least with respect to the core function (hashing) of the networks, both are for now about equally secure.

Number of full nodes

The number of full nodes is sometimes viewed as a measure of decentralization of the network. The reason some Bitcoin developers are averse to a block size increase is they fear it will increase the resources required to run a full node, resulting in fewer full nodes and more centralization. Currently, Ethereum has more than twice the number of full nodes as Bitcoin does, even with bigger block payloads per unit time.

Transaction fees

Other things equal (e.g., mining reward), it is better for a cryptocurrency to have lower transaction fees. Ethereum has dramatically lower transaction fees than Bitcoin does, despite processing more transactions and having an equal overall mining reward (which leads to overall security). Even so, in my opinion Ethereum needs to do the work to ensure on-chain scaling continues so that transaction fees remain low.

Block frequency

Bitcoin processes blocks according to a Poisson process with a mean frequency of 10 minutes. Ethereum’s mean frequency is on the order of 15 seconds. This means you can have some certainty that your transaction will clear much sooner with Ethereum than you can with Bitcoin.

Robustness of scripting language

Ethereum features a Turing-complete instruction set with several normal-ish, expressive programming languages, while Bitcoin has only limited, gobbledygook op codes. This means that Bitcoin can only represent a ledger or some very basic conditional payments, while Ethereum can represent an entire computer state. I say this is a clear advantage for Ethereum, while some critics try to twist it into a point for Bitcoin. I will address the critics in the next section.

Governance quality

Bitcoin’s governance quality is abysmal, as discussed above. Meanwhile, the Ethereum community is practical and congenial. Both teams have talented computer scientists, but Ethereum’s culture is infinitely better. I am consistently impressed with Ethereum founder Vitalik Buterin’s maturity, humility, and leadership skills. Sometimes Vitalik is absurdly criticized as a dictator, but this accusation has no basis in reality. He seems to intuitively understand the need to build a broad consensus within the community, and this consensus in fact exists.

Sometimes Ethereum’s governance institutions are criticized for the 2016 bailout of the DAO investors. I will address the critics in the next section.

Final score

Bitcoin continues to have a clear lead in market cap (network effects) and trading volume (liquidity), and it is tied with Ethereum in block rewards (chain security). Ethereum leads in every other dimension: number of transactions, number of full nodes, lower transaction fees, block frequency, scripting capability, and governance quality.

If Ethereum were to appreciate enough to pull ahead in market cap (the “flippening”), it would also pull ahead in block rewards. And it’s hard to imagine that liquidity would be far behind. In other words, if ether doubled in price and bitcoin halved in price, Ethereum could pull permanently ahead of Bitcoin on every or almost every dimension that matters. This seems not only plausible to me, but likely.

Many of the criticisms of Ethereum are overstated or even unhinged

Bitcoin absolutists often spout criticisms of Ethereum that I will partially address below. The criticisms vary in terms of plausibility, and they are all repeated with tribalist fervor. Let’s take a closer look.

Ethereum has a greater attack surface

Ethereum’s instruction set is Turing-complete, while Bitcoin’s is not. This means that you can write complex smart contracts on Ethereum, and you generally cannot on Bitcoin. Bitcoin does whitelist some basic smart contracts as op codes, but it will probably never have as robust capabilities as Ethereum.

The other side of this coin is a concern raised about security. The fact that you can do more on Ethereum leads to some challenges, which are often called “a greater attack surface” by critics. There are two ways this criticism could be valid, one of which seems not to matter in practice at this time and one which isn’t a fair apples-to-apples comparison.

The first is related to setting the gas cost of operations on the Ethereum network. If some operations are underpriced in terms of gas relative to their computational cost, then it becomes possible to flood the network with underpriced but computationally costly operations. This brings legitimate activity to a halt, a denial-of-service attack. These kinds of attacks occurred several times in 2016, resulting in a repricing of certain operations. Presumably, as underlying computing technology changes, operations will need to be repriced again in the future. But this kind of attack only denies access to the network until gas values are recalibrated; no coins are at risk. Once recalibration is done, the issue basically goes away. While this is an issue the community needs to remain on top of as computing capabilities change in the future, it’s not a serious concern right now.

The other way in which Ethereum has a larger attack surface relates to individual smart contracts. With a more robust programming language, there are more ways to screw up. Bugs in complex code lead to debacles like the DAO fiasco. Yet, on an apples-to-apples basis, Ethereum transactions are every bit as secure as Bitcoin’s are. Bitcoin does not allow the kinds of complex smart contracts that are likely to have some of these bugs. If your goal is only to do the kinds of transfers on Ethereum that are allowed on Bitcoin, there is no reason to believe that the attack surface is any bigger. The code to transfer value on Ethereum is even simpler than it is on Bitcoin.

It is true that irrational exuberance about smart contracts on Ethereum has led to some projects launching poorly reviewed code that was ultimately insecure. But for the kinds of applications you might use Bitcoin for, Ethereum appears to be as secure as Bitcoin is.

What about the DAO fork?

In general, my prior is to oppose bailouts, and there is no denying that the hard fork that followed the DAO debacle was a bailout. Although I did not actively support the bailout at the time, I’ve arrived at a relatively agnostic perspective. Looking only at the consequences, it’s clear that the sky did not fall because the DAO investors were bailed out. Ethereum has proceeded quite successfully since July 2016. It’s possible that ether would be worth even more today had they not bailed out the DAO investors. It’s difficult to know. I would say that it is a plausible view, by no means certain.

What is an implausible view is the continual carping about bailouts on Ethereum, the implication being that the Ethereum community will hard fork to bail out everyone else who is careless with their smart contracts. Even if you think the DAO hard fork was misguided or wrong, there are numerous differentiating factors that make it separable from current activity. First, at the time of the DAO attack, Ethereum was still a young, immature system, less than a year old. Second, the DAO held approximately 15 percent of all ether that had been issued up to that point. Nobody wants an attacker to hold 15 percent of all tokens on any chain.

Finally, in recent coin loss events (like the Parity multisig bug), virtually nobody proposed hard forking Ethereum. It’s fine to continue to oppose the DAO hard fork that occurred 18 months ago. It’s unhinged and dishonest to pretend that Ethereum is bailout central for all time.

Vitalik is a dictator

Bitcoin was started by “Satoshi Nakamoto,” a pseudonymous founder who later disappeared. Ethereum was largely spearheaded by Vitalik Buterin, a Russian-Canadian twenty-something who is still around and actively contributing to the project. Vitalik exercises an intellectual leadership role within the community, which in my opinion he completely deserves on the basis of his contributions, temperament, and demonstrated strong judgment.

Some bitcoiners argue that Vitalik’s outsized influence means that Ethereum development is centralized. This is bananas. Vitalik is influential because he has a strong track record. As far as I can tell, technical dissent is extremely well tolerated among the Ethereum development community. And although Vitalik supported the DAO hard fork, to my knowledge he has not so much as said one negative word about those who chose to avoid the fork and continue with Ethereum Classic.

Meanwhile, Bitcoin core development is extremely ideologically centralized (and cynically, one might add centralized around the financial interests of Blockstream). There are purges of technical dissenters. There is widespread censorship on /r/Bitcoin. In contrast, Ethereum development has an extremely tolerant culture with plenty of room for back-and-forth and compromise.

Hard forks are unsafe

It has been an article of faith among some in the Bitcoin community that hard forks—introducing backwards-incompatible changes—are dangerous. Yet, somehow, Ethereum has hard forked on several occasions—mostly to add features—and everything has worked out just fine. How can this be?

Hard forks surely can be unsafe. If they are executed without adequate preparation, testing, notice, or consensus, they can result in chain forks or other debacles. But hard forks are also a necessary tool for adding features to existing blockchains. When executed properly, they do not appear to be unthinkably dangerous—hence Ethereum’s successful track record of hard forking.

One’s view of hard forking appears to depend on where you see blockchains in the development and adoption cycle. Some people think that blockchains are like the early World Wide Web—we’re in the “Netscape days.” I think more accurately we are in a state like the early Internet—we’re in the “ARPAnet days.” We are a loooooong way from reaching the full potential of blockchain technology, and therefore, upgrades to current functionality are vital. Inevitably, these upgrades will be provided through hard forks.

The belief that all new functionality will come through Layer 2 services or through soft forks is wrong. Knocking Ethereum for embracing and successfully executing hard forks is wrong-headed and counterproductive.

The store of value function of Bitcoin is overrated and can be accomplished with a token on top of Ethereum

“OK, sure,” I have heard some people say. “Ethereum is a better medium of exchange, especially since it has lower transaction fees. But Bitcoin is a better store of value.” Bitcoin has a hard cap of 21 million coins that will ever be issued—it is digital gold, the story goes.

This story is a retreat from the original Nakamoto paper, which clearly envisioned Bitcoin as a payment system. But even if we accept this change in scope, it’s not clear a) that Bitcoin can accomplish its store of value mission without also being a medium of exchange, b) that storing value on a blockchain with minimal possibility of transferring it is very valuable, or c) that a better store of value could not be achieved by creating a scarce token on top of Ethereum.

Bitcoin security is paid for with mining rewards, which consist of block subsidies and transaction fees. Because Bitcoin has a hard cap of 21 million coins, the block subsidy will go down over time and eventually reach zero. This means that to maintain a given level of security, transactions are necessary and there is a need for fees to be non-trivial. Yet if fees are high, transactions will move to other blockchains. The limit to the mining reward is determined by transaction demand—price × quantity—and therefore blockchain security is also limited by transaction demand. The commitment to have a fixed supply of coins potentially implicates a security parameter of the network, meaning bitcoins might not actually be a great store of value on a network where transactions are discouraged by high fees.

In addition, Bitcoin as “digital goldbuggery” has all the same problems that regular goldbuggery does—it stems from a misunderstanding of monetary economics which holds that all inflation is bad or even theft or fraud. I hold zero real gold—so why would I need digital gold, especially digital gold that is ridiculously expensive to transfer? While there is an interesting possibility for cryptocurrencies to supplant traditional monetary policy, that opportunity is not by fixing supply but by establishing a token that offsets changes in the spending level with increases or decreases in the supply of coins to keep nominal spending constant in a given geography. That is an interesting concept. But gold is dumb, and digital gold is also dumb.

Finally, even accepting arguendo that there is a need for a fixed-supply digital token, it’s not clear why that mission could not be better accomplished by establishing a fixed-supply token on top of Ethereum. You could have a token that has a supply of 21 million, that can be transferred for Ethereum’s lower network fees and with its faster blocks, and that takes advantage of its built-in zk-SNARK capability so that transfers are not traceable. Such a token would dominate Bitcoin even as a “digital gold” store of value. Historically, you might have argued that Ethereum’s mining rewards were lower than Bitcoin’s, and consequently, such a system would not be as secure as Bitcoin. But as discussed above, that is no longer the case. Therefore, the continued existence of Bitcoin is unnecessary even by the terms set out by digital goldbugs.

The Ethereum ecosystem is developing rapidly

As I wrote above, we are still in the early days of cryptocurrency technology. The future belongs to the ecosystem with the most rapid development and maturation trajectory, as well as the most willingness to safely adopt changes to the core network to support these advances. Here are a few areas where I think Ethereum is leading.

Scaling

Most of Bitcoin’s scaling effort has gone into adopting SegWit (which reduces on-chain transaction size) and enables Lightning as a level-2 solution. Ethereum also has a Lightning-like level-2 solution known as Raiden. Raiden will work not just with ether, but with any ERC20- (or successor-) compatible token. In addition, developers are working on Plasma, which I understand to be like Raiden but allowing not just transfers but generic state changes, doing for smart contracts generally what Lightning and Raiden do for payments.

In addition, the Ethereum community is exploring additional scaling solutions like proof-of-stake and sharding. You may or may not believe these solutions will work. Presumably, they will be fully tested on secondary networks before implemented on the Ethereum mainnet. If they don’t work, then Ethereum will be no worse off than other coins. If they do work as expected, Ethereum will have a major scaling advantage.

Subprotocols

Some of the most exciting parts of the Ethereum ecosystem have little to do with payments. For example, the Ethereum Name Service is an alternative to the existing Domain Name System. At a time when sites like Sci-Hub are being brought down through the centralized DNS, decentralized ENS has the potential to be a killer app.

Swarm is an Ethereum-native content-based addressing system like IPFS. If a critical mass were to use Ethereum for ENS access, they could also easily have access to decentralized storage and content delivery through Swarm. A site like Sci-Hub could even go without a webhost!

Unlike Bitcoin, which treats SPV clients as an afterthought, Ethereum is doing a lot of work to enable light clients. This will allow mobile clients like Status to exist. Status makes use of the entire Ethereum ecosystem, including ENS and Swarm, and also the Whisper messaging protocol.

Ethereum also has the most advanced token standards. People have been talking about “colored coins” in the Bitcoin community for years, but no one has really made them work. On the other hand, of the top 30 tokens listed on coinmarketcap.com, 28 are on Ethereum. While ERC20 is the basic standard, the community is working on extending the standard in backward-compatible ways.

In addition, work is being done to enable atomic swaps of ether and ERC20 tokens on the Ethereum blockchain. This will enable fully decentralized exchanges that can solve a major headache associated with the cryptocurrency ecosystem, the security of centralized exchanges (remember Mt. Gox?).

Useful apps

One of the most exciting applications that can be potentially achieved with Ethereum but can’t with Bitcoin is decentralized prediction markets. The Ethereum ecosystem features two such apps, Gnosis and Augur. Prediction markets have the potential to change the world.

Importantly, prediction markets could even be used to help make decisions regarding cryptocurrency governance. Imagine that the DAO heist happened in the presence of a fully functional combinatorial prediction market. The community could have launched a prediction market for the price of ether at a particular date conditional on there being or not being a hard fork to reverse the heist. The community could take the path most likely to raise the price of ether. Such Futarchy is a fascinating idea, but there’s only one blockchain that is likely to support it any time soon.

Bonus section: Why Ethereum will beat Ripple

Ripple has been on a tear lately, and now ranks as the number two cryptocurrency by market cap. As far as I can tell, Ripple is feature-dominated by Raiden, Ethereum’s level-2 token transfer network. Ripple is not a fully open network, and although it can transfer tokens representing any assets, the vast majority of tokens in use today live natively on the Ethereum platform. As such, it seems natural to transfer them using Raiden rather than establishing secondary tokens on Ripple to transfer a representation of Ethereum tokens.

Ripple’s high valuation makes Ripple co-founder Chris Larsen one of the richest people in the world on paper. CNBC calculated his notional net worth at $59.9 billion, richer than Larry Ellison, Larry Page, and Sergey Brin. Given the actual level of Ripple usage, this seems likely to end in tears for XRP investors.

Maybe not permanent maximalism

For now, I am an Ethereum maximalist. I think it makes sense for the cryptocurrency market to be dominated by Ethereum and tokens built on top of Ethereum. I no longer believe there is a stable place for Bitcoin, Ripple, or most other cryptocurrencies that exist today.

But in the long run, I am open to the possibility that a handful of other cryptocurrencies could develop and circulate alongside Ethereum in a more-or-less permanent way (or even supplant Ethereum). Given my model of cryptocurrency competition, it is fairly clear how such currencies could operate—they could be better on specific, useful margins than Ethereum is, or they could be optimally designed for specific applications for which Ethereum is a poor fit. (Or, given the latencies involved, perhaps Mars will have a separate cryptocurrency.)

I will always be grateful to Bitcoin for what it represents—a revolutionary advance in computer science. But for now, the world has moved on. Markets just haven’t recognized that yet.