Will network effects save Ethereum from the Ethereum killers?
First-generation, established blockchains, like Bitcoin and especially Ethereum, are facing competition from an increasing number of newer and faster layer-1 blockchains like Avalanche, Cardano and Solana. In this context, it is useful to carefully assess the network effects of Bitcoin and Ethereum in order to decide the extent to which they can provide defensibility against newer blockchains.
Recall that a product or service exhibits network effects when the value to a user increases in the number of other users that buy the same product or use the same service. How does this definition apply to blockchains?
To answer this, it is useful to first note that there are three main types of constituents for layer-1 blockchains like Ethereum: validators (miners for proof of work or stakers for proof of stake), developers and end-users. In principle, there can be same-side network effects among the members of each of these three groups, as well as cross-side network effects between any pair of groups – so six total possible sources of network effects. We consider each in turn.
Ethereum’s network effects
a) Same-side network effect between users
End-users “adopt” a blockchain by opening up a wallet denominated in that blockchain’s token (cryptocurrency). So for end-users, adopting a blockchain is synonymous to adopting that blockchain’s token. And the more people have wallets corresponding to a specific blockchain’s token, the more attractive that token becomes as a medium of exchange (e.g. buying goods/services with that token from sellers who accept it). This is of course the classic network effect enjoyed by any currency.
However, once the number of users and transactions on a blockchain becomes large enough, the same-side network effect between users can turn negative because of congestion. This happens sooner on proof-of-work blockchains like Bitcoin and Ethereum. The more users conduct transactions on a proof-of-work blockchain, the slower the entire network, because it takes longer on average for transactions to get validated.
Another issue with the same-side network effect among users is that cryptocurrency wallet providers (e.g. Coinbase) are making it increasingly easier to hold and transact many different cryptocurrencies, i.e. to “multihome” across many blockchains.
b) Same-side network effect between developers
The more developers build projects/applications on top of any given blockchain (e.g. dapps on Ethereum), the more tools, software, modules and expertise become available for building for that blockchain (e.g. on GitHub/Ethereum), and so the easier it is to develop new applications for it, which attracts more developers to do so. This is very similar to the network effects benefitting programming languages like C++ or Python.
c) Same-side network effect between validators
Validators compete to validate transactions, so there is a straightforward negative effect between them that takes the form of business competition. Everything else equal, each validator would like to face less such competition.
d) Cross-side network effect between developers and users
Higher user demand for a blockchain makes developers more willing to invest in developing apps because it signals to them a larger pool of potential users for their apps. And a wider selection of apps attracts more people to adopt that blockchain.
However, it is worth pointing out that this effect is significantly weaker than the network effect between users and developers on a centralized platform like Apple’s iOS. This is because, unlike the case of iOS with its centralized App Store, developers do not get discovered by users by virtue of adopting a popular blockchain like Ethereum.
Furthermore, just like in (a), once the number of developers on a blockchain becomes large enough and induces a sufficiently high amount of usage, the network effect turns negative due to congestion. At scale, congestion can be significant and overwhelm the positive network effect: for much of 2021 to date, as Ethereum experienced much wider adoption and usage, congestion issues seemed to dominate any positive benefits coming from this greater usage. This has driven some developers (and users) to consider alternative blockchains built on proof-of-stake (e.g. Solana). Indeed, the main advantage that blockchains based on proof-of-stake consensus have over the ones based on proof-of-work is speed (significantly less congestion), albeit arguably with some sacrifices in terms of decentralization.
And there is one more problem with the users-developers cross-side network effect for blockchains. To the extent that users tend to follow blockchain-based apps wherever they are, they will be willing to adopt the blockchain adopted by the developer. Yes, that may require users to add a new token to their wallets, but that is a small friction in most cases, especially with wallet providers making it increasingly easier to hold multiple tokens. This means there is no real coordination problem of switching blockchains if developers find an alternative blockchain is better.
e) Cross-side network effect between users and validators
All other things equal, adding more validators to any given layer-1 blockchain (miners for proof of work or stakers for proof of stake) makes it safer for users to hold its token and engage in transactions using it. The increase in safety comes from the fact that it is more difficult for any individual validator to manipulate transactions when there are more other validators. Conversely, the more users hold tokens and engage in transactions, the more validators will be attracted since there are more revenue opportunities from validating transactions.
This network effect is important in the early stages of a new blockchain, but from the perspective of users, it arguably plateaus fairly quickly, once a critical number of validators and/or corresponding hashing power has been achieved.
f) Cross-side network effect between developers and validators
Similarly to e) above, adding more validators also makes any given layer-1 blockchain safer for developers to build applications on top of it. And the more applications are developed for a blockchain, the more attractive it becomes to validators.
Also, similarly to e), this network effect probably runs out of steam quite quickly for developers, and so does not provide much long-term defensibility.
Price effects are not network effects
One possible confusion around the network effects of a blockchain is the fact that users, validators and developers holding a particular blockchain’s token stand to benefit financially when more other participants come onto the blockchain. This pushes up the token’s price given supply is limited. But an increase in the price of the token doesn’t necessarily mean the blockchain generates more utility for participants, which is what is relevant for it to have a network effect.
Undoubtedly, the price of a given blockchain’s tokens serves important roles, including acting as a signal of the blockchain’s popularity. However, creating a network effect is not one of them. Saying a blockchain has a positive network effect because its token price goes up as that blockchain attracts more users would be like saying coffee has a network effect because its price goes up as more people want to drink coffee. That is just a normal demand-supply effect, and not a network effect. This is why any claim of network effects should focus on utility, taking prices as given. Even if the reason for the price increase stems from speculative beliefs rather than an increase in real demand, this does not create a network effect. Otherwise, we would have to say that fine art, stocks and tulips all enjoy network effects, which is obviously silly. Higher expected prices increase the value of holding an asset, not the value of consuming it. The former does not offer defensibility for the underlying product, which is the whole point of network effects.
Concluding thoughts
Understanding the various network effects surrounding blockchains is not straightforward, and it is easy to conflate them with demand effects and the role of self-fulfilling beliefs in asset pricing. This is why we thought it useful to disentangle and discuss them in this post.
So how defensible are the network effects described above? Put differently, how confident can we be that Ethereum is protected by a first-mover advantage?
To us, out of the six network effects discussed above, the strongest and most defensible for Ethereum right now is the same-side network effect among developers. There are more developers working on Ethereum projects than on any other blockchain. To the extent there are more new Ethereum projects than new projects on any other blockchain, its advantage will continue to grow in this respect.
Overall, therefore, we think Ethereum has quite strong defensibility, at least for the next few years, primarily driven by the network effects arising from its developer ecosystem. The congestion issue that we discussed earlier can be managed to a certain extent by making use of popular layer-2 solutions like Polygon, which make Ethereum more scalable. And it seems we are not too far away now from the long-anticipated launch of Ethereum 2.0 which will involve a switch to proof-of-stake, thereby addressing the congestion concerns head on.
That being said, new layer-1 blockchains like Solana rely on more popular programming languages, which could partially nullify the developer expertise network effect advantage enjoyed by Ethereum. Developers are also increasingly multihoming. More fundamentally, one can imagine a future in which it becomes seamless to make transactions across different competing blockchains via tools that ensure interoperability. This would make network effects work at the level of the entire blockchain economy rather than at the level of any individual blockchain, meaning no particular blockchain would enjoy defensibility, Ethereum included.