The World does not need a World Computer
The World does not need a World Computer
Part II of the case against the Flippening. Ethereum is an expensive and complicated solution to a problem no one actually had.
This is Part II of a two part series carefully examining the investor thesis for owning Ethereum and addressing reader questions about the "Flippening" - the scenario when Ethereum’s market cap overtakes the market cap of Bitcoin. If you haven’t already read Part I: The Trouble with Ethereum I would recommend starting there. As always none of what follows is financial advice - you should never get financial advice from anyone who asks you to like and subscribe.
On a related note, please like and subscribe!
Own the Internet: Ethereum the World Computer
What is a dApp for?
Decentralization as a Service
Ethereum might be too decentralized
Ethereum might not be decentralized enough
The bull case for Ethereum
Own the Internet: Ethereum the World Computer
The most common arguments in favor of investing in Ethereum are that Ethereum is "ultrasound" money (we addressed this argument in Part I) and that owning Ethereum is like owning an index fund comprised of all the things built on top of Ethereum. A good example of this line of reasoning comes from Packy McCormack:
Owning ETH is like owning shares in the internet. Demand for ETH will go up with increased web3 adoption, while upcoming changes will decrease the supply of ETH and let more value accrue to holders. It’s like a tech stock, a bond, a ticket to web3, and money, rolled into one. - Packy McCormack, Own the Internet
Beware of arguments by analogy. Owning Ethereum does not give you any claim to the downstream applications built on top of it and a lot of things need to be true for value built on top of Ethereum (the platform) to trickle down to ETH (the asset). Applications built on top of Ethereum are more like customers than subsidiaries. They are a valuable source of revenue but they don’t belong to Ethereum and they aren’t necessarily loyal.
To unpack this a bit more, let’s consider a hypothetical decentralized application (often known as a dApp), built on Ethereum: d'Uber.1
What is a dApp actually for?
A decentralized application, as the name suggests, is any application built using some kind of decentralized ledger or blockchain. It’s not immediately obvious why someone would want a decentralized application. As we talked about in Part I blockchains are intrinsically expensive tools (Axiom #2) that are only useful for resisting change (Axiom #1). Decentralizing is just a ten dollar word for making something more expensive and harder to update.
Decentralizing something makes sense if (and only if) it is extremely important that the rules of the system can’t change. Using a blockchain is like hiring the world’s most expensive (but also the world’s most honest) referee to enforce your rules. It only makes sense to do when the stakes are very high and the rules are very stable. If the stakes aren’t high there are cheaper solutions and if the rules aren’t stable a blockchain would only get in your way.
Decentralization is expensive and inconvenient, so even the applications that use it seek to use it as little as possible to achieve their goals. That’s why all the major dApps on Ethereum today use smart contracts to enforce their logic but run ordinary centralized web servers to provide the UI most users interact with. It would make no sense to run a website on a blockchain - it would be enormously expensive and nobody really cares if a website gets updated.
So our example app d'Uber would probably still have a website. It’s not immediately obvious what parts of Uber are worth the expense of decentralizing. Presumably the payments between passenger and rider? Probably the pricing logic? Possibly the matching engine? Any part of the system they decentralize will be inherently more expensive and have slower response times. Decentralized Uber will have higher operating costs and worse performance than traditional Uber, and the more decentralized it is the worse that comparison will become.
Axiom #4: Applications will seek to use as little decentralization as possible.
Most applications don’t need any decentralization at all, but even those that do will use it as sparingly as possible to minimize costs.
Decentralization as a Service
As I said above, I think it is better to think of decentralized applications as customers of Ethereum than subsidiaries. dApps pay miners to decentralize certain aspects of their operation by defining them in a smart contract and encasing it in blockchain amber. Ethereum is essentially a service provider and like any service provider its customers are price sensitive and subject to competition.
In our thought experiment above for example users of d'Uber will not care what tech stack is enabling them to hail or offer rides. They will prefer whichever solution is the cheapest and most effective for their needs. d'Uber will decentralize only to the extent its customers demand it and no more (Axiom #4) and it will rely on whatever service provider can sell it that decentralization most cheaply.
Price competition among those service providers will be intense, because networks can imitate each other and make the cost of switching to a competitor arbitrarily low. Most of the major smart contract platforms today (the so-called Ethereum killers) are compatible with the Ethereum Virtual Machine (EVM). That makes it easy to migrate or copy successful Ethereum dApps over to competitor networks. In economic terms, decentralization is a commoditized product.2
Viewed through this lens many of the advantages of the Ethereum network appear more fragile. Ethereum has by far the largest developer community, which has legitimate and powerful network effects. But those network effects support the developer community itself, not the Ethereum network. If it is easy for dApps to migrate to a new network it is easy for the developers and vice versa. Ethereum can be replaced without diminishing the strength of the developer network.
One way to see this is in how many of the most important technological innovations born in Ethereum (Solidity/EVM, ERC-20, ERC-721, etc) are now table-stakes features of every smart contract platform. They grow the ecosystem at large but they don’t provide any competitive edge for Ethereum or ETH investors.
The large pool of capital and developers that DeFi has accumulated is often held up as evidence of Ethereum’s strength, but it is better understood as an ever growing reward for any successful Ethereum competitor. The value of the DeFi market is not the same as the strength of Ethereum’s hold on it. To believe in Ethereum as a world computer you need to believe that DeFi is growing but you also need to believe it will stay on Ethereum. That second part is often taken for granted.3
To be successful Ethereum must walk a knife’s edge between being too decentralized and being not decentralized enough.
Ethereum might be too decentralized
The most visible complaint that Ethereum’s customers have with the network (and hence the obvious target of most competitors) is how expensive it is to use the network - generally referred to as gas fees. At time of writing the 7 day average cost of a transaction on Ethereum is ~0.0074 ETH (~$25.6/transaction). Larger more complicated transactions like the ones used by dApps can easily cost ~20-40x more gas than a typical transaction.
High transaction fees aren’t inherently bad. They prove that users place value on transacting on the network and they help pay for network security. But they do represent a competitive vulnerability, or perhaps even an existential threat.4
Given that d'Uber would run equally well on either Solana or Ethereum, it isn’t clear why d'Uber users would be willing to pay high fees for the Ethereum version. The cost of decentralization is ultimately passed on to the users and users are price-sensitive, so decentralized applications need to be price sensitive as well. It is no surprise that the primary feature offered by most Ethereum competitors is cheaper transactions.
The way that these platforms achieve cheaper transactions is through a combination of fewer users (and hence less traffic congestion) and by compromising on decentralization in various ways. If you think of decentralization as the overarching goal of dApps that makes Ethereum seem obviously superior, but dApps are not seeking to maximize decentralization at all. Their goal is to buy the minimum necessary decentralization at the cheapest possible cost. (Axiom #4)
d'Uber does not care about decentralization as a per-se goal. The actual goal of d'Uber is to enable people to purchase and sell rides. Decentralization is just a means to an end - it’s entirely possible that a compromised but still somewhat decentralized competitor with drastically cheaper fees will be "good enough." It depends on which threat models turn out to be significant.
Bitcoin’s decentralization is extreme because its threat model is extreme - governments go to war over money, so Bitcoin is built with military grade decentralization. Not all applications will need that - for many applications it might enough to have modest or even symbolic levels of decentralization that can be overridden from time to time without compromising the goals of the system. The more decentralized a system is the harder it is to change - but for many dApps being moderately inconvenient to change may actually be enough.5
Governments may also decide to relax regulations and diminish the need for decentralization as a service. Crypto may end up having the same effects on banking and financial regulation that Uber had on taxi regulation or AirBNB had on hotel laws. If government control is less burdensome, dApps will be less willing to pay to escape it. They may migrate to less decentralized platforms or even to platforms that aren’t decentralized at all.
Ethereum might not be decentralized enough
One of the main complaints you will hear about Ethereum from critics is that it is centralized. Part of the accusation here has to do with the arguments around using Ethereum for money that we discussed in Part I. But that criticism also applies to the idea of Ethereum as a "world computer." If the Ethereum system has any chokepoints of central control, then the blockchain it operates is just expensive theater.
If you hear people call Ethereum a scam, this is the complaint they have in mind - Ethereum is charging a high price for a product that may not work as advertised. Ethereum works today but it has not yet sustained any serious attack. The economy built on top of it is like a seaside village on the Japanese coastline - it may thrive for some time before a tsunami washes it away.
Today the main practical use of smart contract platforms is for obviously centralized companies to yell "decentralized!" and then ignore whichever regulations they find inconvenient. It's like yelling "international waters!" as you open a houseboat casino 10 ft from shore. I’m honestly surprised it has worked as well as it has, but eventually authorities will probably decide there shouldn’t be a secret code word that lets you do crimes. The transition will be edifying.
When the waters rise we will learn which platforms are sufficiently decentralized to resist control and which are decentralized in name only. Ethereum has a history of quick and decisive action as a network - which does imply a decision making authority of some kind. If that’s true, that body could be co-opted by governments or even just corrupted by its own self-interests. If Ethereum isn’t decentralized enough when it counts then decentralization won’t actually provide any benefits.
It is difficult to be precise about exactly how decentralized is decentralized enough until those theories are meaningfully tested - but there are some some fairly visible vulnerabilities you can imagine an attack exploiting. It is difficult to run an Ethereum node so many people rely on the centralized third-party service Infura rather than running one themselves. Even among those who do run their own nodes ~60% of them are run on Amazon Web Servers or other cloud providers. The government could send letters to Amazon and Infura and shut down the majority of Ethereum network.
There are also worries about how much influence founders and early developers have in Ethereum. The Ethereum blockchain once overrode the code-is-law ethos to undo a successful hack of The DAO. Roughly a year later they chose not to do a similar rollback to rescue funds frozen in a Parity bug. It is possible this was the spontaneous, organic will of the community. But it is notable that Ethereum founders stood to lose millions in the DAO hack but were largely unaffected by the Parity bug. Even more recent choices like the launch of EIP-1559 are deceptively political.
The developer team being able to change Ethereum at will is problematic even if they are trustworthy because the whole point of decentralization is to eliminate the need for trust at all. A rumor about Vitalik’s death once shaved ~$4B off of the Ethereum market cap. He ended up posting a picture to Twitter to prove he was still alive:
If a significant portion of Ethereum’s value comes from Vitalik’s participation, that implies a significant portion of Ethereum investors are effectively trusting Vitalik. What happens if his interests diverge from theirs? What if Vladimir Putin calls him up and demands his cooperation?
It is much easier to build decentralized applications on Ethereum than it is on Bitcoin, but if Ethereum itself can be compromised it doesn’t really matter. Decentralized applications can’t purchase the decentralization they need from a platform that isn’t decentralized.
The bull case for Ethereum
To be excited about investing in Ethereum you need to believe that smart contract usage will continue to grow - but you also need to believe that Ethereum is immune to government intervention and that users want to pay for immunity and that no other network can offer it to them more cheaply or in smaller quantities and that the forces that shape Ethereum’s political choices can be trusted to remain benevolent.
Even if we knew for sure the future of finance was being prototyped on Ethereum today that wouldn’t necessarily make ETH a good investment. Owning a token is not the same thing as owning the efforts of everyone who is interested in the problems your token was meant to solve. Building the future is not the same thing as profiting from it. Too many Ethereum investors are glib about the difference.
So to answer the original reader question that prompted this essay:
"I'm seeing a lot of talk these days about a flippening between BTC and ETH. Is there a scenario in which you ever see that happening?"- SH
No.
As we will see, Uber is actually a very bad choice of application to decentralize - which is helpful in terms of understanding the drawbacks of using a blockchain. The same arguments apply in other examples such as comparing a decentralized exchange to Coinbase or comparing a DeFi lending protocol to a traditional bank - but I think Uber is easier for non-traders to form intuitions about.
There is one exception to this, which is that all networks have a natural monopoly on their own native token. Other networks cannot undercut the price of moving ETH because Ethereum is the only network capable of moving ETH. That’s why we discussed the argument for using Ethereum as money separately in Part I.
That’s not to say that Ethereum has no advantages over nascent "Eth killers" - but that advantage doesn’t come from the size of their economy it comes from how long it has existed. Most high-end NFT holders still prefer to collect NFTs on L1 Ethereum in spite of the fees is because L1 Ethereum has been around the longest of any of the smart contract platforms and is therefore the most Lindy. NFTs work just as well on any platform
The most common answer to this concern is that Layer 2 technologies like Polygon or Arbitrum will reduce transaction fees - but Layer 2 technologies will be equally useful on any EVM-compatible platform. They will benefit the smart contract ecosystem but they will be too easily co-opted to be a competitive advantage for Ethereum. Layer 2s are better thought of as purchasing unions for smart contract customers - they serve the interests of their own customers (dApps), not the interests of smart contract platforms.
A good example is the Liquid Network a sidechain of Bitcoin that is run by a committee of major exchanges. You can use the Liquid Network to manage your Bitcoins and it is much faster, cheaper and more private than the underlying Bitcoin network. The downside is that you have to trust a majority of the exchanges that run it. If the alternative was just keeping the Bitcoin custodied on one of those exchanges anyway, that’s a big improvement even though it is definitely not full decentralization.