EOS: Don’t Believe The Hype
Before I start, I would like to disclose that my crypto asset portfolio is heavily weighted towards $ETH and that I do not own, nor have ever owned, $EOS.
I am incredibly conscious that, while I try and remain as objective as possible in my analysis, my financial stake will inevitably result in some degree of confirmation bias on my behalf.
I encourage anyone reading this to refute my assertions and help me understand the flaws in my perspective.
As always, the following is for educational purposes only. This is not investment advice.
All references to spot price and market cap are relevant as of 04/29/18.
EOS, a smart-contract platform developed by Steem and Bitshares founder Dan Larimer, has received a lot of attention in the run up to its mainnet launch in June.
This attention has been directly reflected in the project’s remarkable market activity: $EOS, with a market capitalization of close to $17bn, is now valued higher than both Litecoin and Cardano combined; $EOS’ daily volume is double that of the second largest crypto asset, Ether ($ETH); and $EOS’s price has continued to trade strongly against both market leaders Bitcoin ($BTC) and $ETH.
I have had reservations regarding EOS since the project’s inception. In the wake of the Multicoin Capital ‘Analysis & Valuation’ report last week I thought it would be appropriate to air these reservations.
While I am often an admirer of the content issued by the Multicoin team, I find their EOS report to be overly biased and ignorant as to the very value that blockchain technology, through a combination of cryptography and economic incentives, provides.
In order to understand EOS’ design principles, it is important to understand the basic scalability trilemma at the heart of blockchain design.
As initially proposed by Vitalik Buterin and further explored by Multicoin, blockchains suffer from a trilemma in that they can currently only simultaneously achieve two of the following three properties:
- Scalability — a high number of transactions per second.
- Decentralization — a large number of actors participating in block production and validation process.
- Security — making it expensive to gain majority control over the network.
For some context, Bitcoin’s consensus mechanism, Proof of Work (PoW), (theoretically) favours decentralization and security over scalability: BTC is permissionless, so anyone can participate in mining or run a node and the security of the network is protected by the price of energy/hardware needed to produce the accumulated hash power.
(In reality, the proliferation of ASICs, specialized and expensive hardware for $BTC mining, has meant that hashing power is largely concentrated in the hands of a few entities.)
Ethereum, the second largest network by market cap, similarly favours decentralization and security over scalability.
It’s current consensus mechanism, PoW, is identical to that of BTC, and its future Proof of Stake (PoS) mechanism will similarly be designed so that anyone will be able to participate in block production: indeed, Ethereum’s PoS may arguably be better than BTC’s PoW in that stakers will not be required to purchase expensive hardware to contribute to the block production process. As a result, the distribution of block producers should be more varied and thus the network further decentralized.
Decentralization over scalability:
Why have BTC and ETH favoured decentralization over scalability?
Commentators may justifiably argue that in order to ultimately be successful, in order to see mainstream adoption of these platforms, these networks must be able to compete with Visa-like throughput. No one wants to use applications in which interactions take days, hours, or even a handful of seconds.
This is a valid critique of blockchains, and yet it fails to recognize the fundamental value that blockchains provide: censorship resistance. If blockchain platforms do not provide censorship resistance — i.e. they rely on a set number of trusted actors to produce and validate blocks — then we have simply returned to legacy database systems, albeit at the expense of the efficiency that legacy systems like Amazon Web Services provide.
Why is censorship resistance important?
The censorship resistant nature of the Bitcoin blockchain means that anyone can hold value in $BTC without the risk of it being seized by a malicious actor, like a government. Eric Meltzer of INBlockchain captures the value of censorship resistance succinctly in the following quote:
“As a jew, I find the existence of basically unseizable money outside of state control massively comforting. To jews, the state is an unpredictable monster that flips out every century or so. Having the ability to leave without leaving everything behind is key.”
Of course, censorship resistance has valuable applications beyond money: Vitalik Buterin recognized this through his creation of Ethereum.
Ethereum allows developers to build censorship resistant applications and contracts. If you enter into a financial contract with another party it is important to be comfortable knowing that neither the record of the contract nor the terms can ever be tampered with at your expense. If your identity — social security number, birth record, driver’s license, credit information, home ownership etc. — exists on a blockchain, it is vital that you can be certain that a set of actors can never interfere with your information. The stakes are clearly very high.
Delegated Proof of Stake:
The scalability trilemma helps provide context as to the design decisions at the heart of EOS. EOS’s consensus mechanism, Delegated Proof of Stake (DPoS) consciously makes a tradeoff between scalability and decentralization.
In the EOS system, a set of 21 Block Producers (BPs) produce and validate each block. EOS token holders vote for these BPs on a 1 token:1 vote basis. BPs are incentivized by validation rewards, which are subsidized by a 5% annual inflation rate.
The concentration of BPs allows EOS to achieve levels of throughput that dwarf the 15 transactions/second (tps) achieved by ETH and the 3 tps by BTC: this is made possible because block information only has to propagate across a set of 21 nodes vs. the 18,000 nodes in the Ethereum network. (For some context, someone achieved 1.5m tps on Ethereum in a similarly permissioned environment.)
EOS’ high level of throughput might seem amazing until you recognize that the system relies on just 21 trusted BPs. Proponents of EOS argue that this is not a problem for two reasons:
- A set of 21 BPs produces a level of decentralization sufficient to allow for ‘platform-grade censorship’. Developers may be uncomfortable building on centralized platforms like Facebook or Twitter knowing that these entities could change the rules at any moment with no recourse. A ‘platform-grade censorship’ platform means that the rug cannot be pulled from your feet at any moment.
- Voters (token holders) can vote to remove a malicious BP at any time. Because BPs have to stake EOS in order to be BPs and are rewarded for their work, they are financially disincentivized from acting maliciously.
These two arguments are inextricably intertwined and do not stand up to critical thinking. The problem ultimately lies with point 2 — on-chain voting systems.
Flaws of on-chain voting:
On-chain voting systems, which attempt to replicate the conditions of representative democracies, are flawed. Vitalik and Haseeb Qureshi both wrote fantastic articles detailing exactly why — here is a summary:
Because blockchains are designed to be permissionless — meaning any anonymous/pseudonymous party can participate in a combination of usage, validation, and block production — networks cannot solve for Sybil attacks, attacks where a single user creates multiple identities to use a network. As such, the 1 man:1 vote system of representative democracies is replaced by a 1 token:1 vote system.
Unsurprisingly, these systems invariably devolve into plutocracies through a combination of both those with more capital disproportionately outvoting their less wealthy peers and those with less voting power becoming apathetic, knowing that their vote will ultimately have little to no effect on the outcome of each election.
Voters therefore become receptive to bribes from BPs, who can buy votes in return for some share of their annual rewards. BPs are encouraged to collude amongst each other so that they can fix the rate at which they will share rewards with voters. If they don’t collude then there will be a race to the bottom, whereby BP profitability will eventually be driven to the equilibrium cost of running the hardware required to be a BP.
Understanding this, it seems clear that the term ‘platform-grade censorship’ is in fact intellectually inane.
If a system has been designed that ultimately encourages plutocracy and collusion among BPs then there is absolutely no guarantee for developers and users that their applications and transactions will not be censored.
Contrary to the myth propagated by certain actors in the blockchain/crypto asset industry right now, censorship resistance does not exist on a dynamic scale: it is very much a binary property.
Because EOS cannot guarantee censorship resistance, it should be viewed as a centralized system, without the throughput advantages that openly centralized networks like AWS provide. Spencer Bogart of Blockchain Capital writes well on this.
Solving the scalability trilemma:
So what is the alternative? As we have discussed, scalability is ultimately required if these platforms are to actually reach a mainstream audience and realize all its revolutionary promises.
The solution is to actually solve the scalability trilemma rather than skirting around it.
In its current inception, Ethereum is arguably centralized, as there are 3 mining pools that together control more than 50% of hash power. However, Ethereum is directly addressing this through the implementation of Proof of Stake, which will have lower barriers to becoming a BP than PoW.
It is worth reading more on all of these solutions. I am not suggesting that any of them are close to being integrated.
The solution that I am particularly excited about and is most relevant to this discussion is PC.
PC refers to child chains in the form of smart contracts. Like state channels, PC chains are ultimately secured by the main Ethereum chain — in the case of any disputes within PC chain, all participants can peacefully exit with their coins to the main chain.
PC allows developers to use any consensus mechanism of their choice, whether that be PoS, PoW, or DPoS, within the child chain. In fact, PC even allows centralized networks to spawn their own chains, meaning they can take advantage of their existing throughput advantages while still allowing users to maintain custody of their funds at all times. A relevant application would be centralized exchanges like Coinbase, Binance, and Gemini.
As a slight digression, with PC in mind, I now envisage a world where all chains exist as Ethereum PC chains, relying on the ever-increasing security of the Ethereum PoS chain.
EOS beyond lack of censorship resistance:
Censorship resistance, or its lack of, is at the heart of EOS’ flaws. I would now like to address several other features of EOS that I think are either unwarranted or overhyped.
In their research report, Multicoin Capital touts EOS’ lack of user fees as a core advantage over systems like BTC and ETH, which require users to pay fees on each transaction.
This idea of ‘zero user-fees’ is disingenuous because it suggests that economic incentives are no longer required to maintain the network. Fees, whether in the form of inflation through block rewards, transaction fees, or a hybrid transaction fee/block reward model, are ultimately what allows blockchains to exist: they incentivize economically rational actors to continue validating and maintaining the state of the blockchain.
EOS recognizes this. EOS does indeed have fees, but they are hidden in the form of inflation. This is not necessarily a bad thing in itself: indeed, both $BTC and $ETH are inflationary at this point in their life cycles, although both assets will ultimately have their supplies capped and rely instead on transaction fees.
So EOS does have fees in the form of inflation. Yes, their system does mean that users do not have to pay transaction fees each time they interact with an EOS-based application.
However, projects will nonetheless have to own some EOS tokens in order to guarantee some level of network bandwidth, so the on boarding process is not entirely frictionless as perhaps suggested.
Moreover, the Ethereum community is already working on solutions to abstract transaction fees away from the user.
Inflation in the context of Store of Value:
It is hard to justify EOS’ long term valuation when it relies on a 5% inflation rate to subsidize BPs.
As Multicoin has recognized in the past, a coin or token must have Store of Value (SOV) properties in order to avoid suffering from near-infinite velocity.
Whether the Medium of Exchange Theory (MV = PQ) is the only way of valuing all crypto assets or not is irrelevant: it is fairly obvious that for an asset to hold value in the long term people must be willing to see it as a suitable reserve currency.
Assets that do not see their value depreciate on an annual basis will be better candidates for the title of SoV than assets with fixed inflation schedules. I recognize that inflationary assets that have utility can also claim SoV status — the United States Dollar is a prime example.
However, the range of these utility SoV assets is limited and I imagine that the crypto asset markets will eventually converge around just a few. Those assets like ETH that have both utility and SoV properties are likely contenders for dominant SoV status.
On the topic of valuation, I think that it is difficult to justify EOS’ current market capitalization and I expect there is some foul play at hand.
At the time of writing $EOS sits at just under $17bn market capitalization, despite having not yet launched a mainnet. For context, it took 21 months of mainnet deployment for $ETH to reach $17bn market cap and $BTC almost 9 years.
It is difficult to justify any valuation in the crypto markets at this point in the asset class’ life cycle but I think that comparable price analysis still holds up. Ethereum was processing over 100,000 transactions per day when $ETH hit $17bn. $EOS has yet to process a single transaction, let alone prove that its consensus model can withstand centralization.
EOS has reportedly raised over $2bn from its year-long ICO. Many analysts suspect that a proportion of funds raised have been recycled back into the ICO in order to inflate the price of $EOS tokens and to signal strength.
Even if these assertions are false, I find it hard to justify such a protracted fundraise, which overtly screams of greed. It does not cost $2bn+ to launch a blockchain. For context, Bitcoin never raised money and Ethereum raised $18m.
Another point of concern is that the $EOS tokens explicitly do not give the token holder any right to access once the EOS chain has been launched. The expectation is that BPs will honour token distribution but there are groups of prospective BPs that have said that they will instead distribute EOS native assets in an equitable manner through an airdrop.
Going back to the discussion of plutocratic rule, I think there is a viable scenario whereby prospective BPs collude to not fully honour the ICO distribution and instead impose some kind of BP token tax. This would actually be possible considering there is a relatively small number of eligible BPs due to the significant hardware requirements needed to qualify.
Developer mindshare and ecosystem funds:
Block.One, the team building the EOS software, will set aside a portion of the funds raised to give as grants to projects building on top of their platform. This is simply not a sustainable growth strategy. Funds are likely to be allocated inefficiently as projects are overly incentivized by grants over actually producing quality applications. Moreover, there is nothing stopping these mercenary-type developers from leaving for the next platform offering generous grants.
A better way of attracting developers is through a strong vision. To date, Bitcoin, Ethereum, Monero and a handful of other blockchains have successfully achieved this, attracting thousands of developers motivated by the idea of building censorship resistant applications.
While these projects did not explicitly give out grants to developers, they did reward early adopters with significant price appreciation of their platform-native digital asset. This will be difficult to achieve with EOS, where the native token is valued at $17bn pre-launch. To achieve returns comparable to $ETH’s, the EOS network would have to be valued at close to $40qd.
Usernames and private key restoration:
Another advantage of EOS touted by the Multicoin team is the username and private key restoration features.
EOS allows users to create human-friendly addresses. On EOS, my address might be @matteoleibowitz, whereas my Ethereum address is 0x23443B1b43E437d6De1fD0v0Za7K3eb0937Y354a.
While these username-wallets are certainly more user-friendly and will probably reduce the rate of mistaken transactions, they significantly reduce user privacy.
In fact, I think that one of the key issues at stake for blockchains right now is lack of privacy. I imagine that any reader would not want to live in a world where anyone could simply search their name on a block explorer website like Etherscan and see how much money is contained within that address.
If that is not the case, then you are always welcome to register an Ethereum-based username using Ethereum Name Service, which has been running for over a year.
EOS will also introduce protocol-level account recovery, whereby a user can specify one or more ‘account recovery partners’ and work with their account partner to reset their owner private key.
However, this only works in situations where private keys are stolen, as the owner needs to use a private key that was active in the last 30 days plus the approval of their recovery partner to reset the private key. This does not protect against the non-nefarious loss/misplacement of private keys, which is likely to be a more common reason for the need to recover account access.
State of the ecosystem:
My last gripe with EOS is that its decision to optimize for scalability incorrectly suggests that scalability, or the lack of, is currently the sole issue affecting the ability of decentralized applications to come to market. I would argue that these applications are still within the conception phase: at best they are in the building phase.
As William Mougayar captures in his State of Tokens presentation (slide 34), we are still very much in the Installation phase of blockchains’ evolution: Vision Realization is only expected to materialize in 2021+.
What about Crypto Kitties, you may ask? Yes, Crypto Kitties did expose the scalability limitations of Ethereum. But I contend that anyone familiar with Ethereum was in no way surprised by this.
Every core developer is publicly cognisant that Ethereum is still in its infant stages. In fact, I think the excitement around the Crypto Kitties saga was more due to surprise that a dApp had managed to secure such popularity at this stage in Ethereum’s life cycle rather than disappointment that the underlying technology was not capable of handling user demand.
Indeed, if scalability were such a make-or-break issue at this point, we might expect to see a backlog of dApps planning on building on EOS. If you look at these two lists (sent my way by two separate EOS evangelists) you will see that that demand simply does not exist. The brevity of these lists is magnified when you compare it to the number of projects building on Ethereum.
Sign up to Crypto Chat, a weekly industry newsletter, here.