In defense of Ethereum and its fatness: a discussion of ETH's value capture potential

Hi everyone,

ETH’s recent drop in price has coincided with a more worrying loss of faith in the so-called “Fat Protocol” thesis which drove the ETH narrative over the past 3 years. A lot of smart people have been questioning Ethereum (and other protocols’) “fatness” (i.e. its value capturing ability) and arguing it is actually “thin”, meaning that even if it succeeds as a network, it will be unable to capture the value it creates.

Crucially, this argument has gathered support from very smart minds in the crypto hedge fund space, including John Pfeffer, James Kilroe, Travis Kling, Gregory Rocco and quite a few others. The idea was also presented earlier this year in Tetras Capital’s 30 page report on why they were short ETH. Since then, ETH shorts have hit record highs and the price has dropped over 50%, with TechCrunch declaring “the collapse of ETH is inevitable”.

In this article, I present and break down the main arguments for “Thin Protocols”, namely: (1) Economic abstraction (2) Commoditization of the protocol layer (i.e. “Race to Zero” argument) (3) High velocity. I then argue why, in each case, they do not really apply to Ethereum as it currently stands and apply even less to an Ethereum with Casper/PoS.

Would really appreciate any thoughts or feedback you might have on the subject.


Thank you for suggesting such a great topic and it’s a good time for discussion in this moment facing ethereum 2.0.

I’ve always wondering i) how could we capture the value of ETH and ii) whether this is really an issue for ethereum itself.

[ TL;DR ]
Regarding your opinion, I DO think,

i) the thin protocol is not that convincing.
ii) For mass adoption of ETH and ethereum itself, non-value capture would not be the crucial problem.

Could you explain the (2) more? I think this part is more relevant with the traditonal business logics. (race to zero, commodization…) Ethereum is not a traditional business indeed.

IMHO (3) High velocity is not that relevant issue with the weight of protocol. The velocity problem belongs to the economic selfishness of participant in the crypto market.

Um, further, inelastic issuance model in ethereum protocol may be another influencial factor to a certain degree, since it’s not controlled arbitrarily by one authority. But this is just a essential factor of cryptocurrency starting from Bitcoin’s economic philosophy… We could not fix this.


If the main reason of this opinion is the “thin protocol” thesis, it’s hard to agree. The fat protocol of blockchain is “physically” a true idea.

For example, the GHOST protocol makes the chain rule and Casper FFG stablizes the total network, making the blockchain itself. The protocol makes it continual the block creation and linked chain mechanism, and this explains simply the role of protocol in blockchain. The value capture problem is just a kind of ETH valuation in market.

The economic issues are always influenced above the protocol layer, like “relatively thin” smart contract layer. This layer is creating its own world like token contract, dApp, exchanges, etc which is more relevant with the economical issue itself.

The critical point is that much of the economic phenomena is decised outside of the protocol, like a smart contract layer or some political issues, weakly belonging to the blockchain protocol itself.

Maybe the protocol influences the price of cryptoasset(ETH) because of the inelastic supply curve of ETH by the protocol in general, but it cannot be the main reason for justifying Ethereum is a thin protocol. It’s just another category of debate.

Additionally, I disagree with the opinion of tech crunch in this moment. It’s just similar with traditional financial guys’ idea like “the lower valuation, the worse commodity.” But, ETH price is not that representative of ethereum chain’s success.. The main factor of ethereum’s success is the decentralized blockchain protocol and the possibility of real implement of the technical design.

As a platform, the entry barrier must be low. And the barrier of ethereum is the price of ETH. Supposing more network value, more price mechanism, there would be no users of ethereum. (supposing high network value and ETH price is $1,000 again)

Buyers do not prefer the volatile unit pricing model using information goods. This has been verified through several thesis. That’s why we’re taking the membership-kind pricing using SW goods or information goods today.

So, in this moment, I wonder, is the lower ETH price is really an issue?

How about getting the ETH valuation more lower and stable(maybe the speculators are going to kill me lol) even if there’s a higher demand and network value?

I think the price per ETH must be gone down more for mass adoption of network or pricing model of ETH should be tuned like fixed fee model… (maybe the optimal price for security issue is to be researched.,)

Twisting our thoughts, when ETH goes down, it makes a lower entry barrier for a normal person to use the ethereum network because of the lower gas fee and ETH buying price.

The price volatility is the main hassle factor of Ethereum and this makes government and traditional men think the cryptoworld is the world of speculation.

Is there any methods with this ? Feedback or debate with my opinion welcomed !

Hi guys, here’s the post for reference:

1 Like

Economically, I think Ethereum is very much in place to achieve full value capture if its native coin is used as the reserved medium of exchange in facilitating worldwide trading of fractional ownership of assets. Anything less than that (i.e. allowing 3rd-party non-native coins to facilitate the trades) would undermine the structure itself as this may open wide the system to be gamed by various rogue agents.

I think the real challenge is not so much about potential value capture, but rather about how the blockchain trilemma (decentralization, security, and scalability) can be well resolved to achieve the desired value capture.

Nevertheless, I am still not aware of any Ethereum-based smart contract operation that does not need ETH to run or pay for gas. So far all the tx that I did involve certain amount of ETH primarily for gas. If there is really such a case possible whereby I do not need ETH for anything both directly and indirectly, then I would like to know what is the rationale for allowing such state to be possible. And what outcome is expected from this.

Rubin’s argument # (2) stated “There is no reason for ETH to be used to pay gas fees, as through economic abstraction any other currencies can be used.” sounds like if I travel to Europe, there is no reason to pay for things with EUR as I should/can use USD/YEN/INR for payment as well. As we know this is not practical.

In Rubin’s # (3) “Rational, independent and self-interested miners will choose to be paid in assets of their own choosing rather than in ETH.”, what exactly would be the reasons why miners would choose to accept non-ETH currencies? If a miner would choose Binance coin for example, in which the coin can be subject to insider manipulation, then how would such falsely inflated value going to affect the network’s trilemma? Did Rubin ever consider such conflict? If so, where is his explanation?

Why would Rubin argue that tokens should have greater value than the coin they are operating on top of? And why would he think that these myriads of tokens would not be less valuable than ETH? What logic is used to justify ETH being worthless while non-ETH tokens being worthy? Any factor that can be applied on ETH to make it worthless should equally be applicable on any token to make that token worthless too. Otherwise, this is obviously to me a case of arguing why the tail of a dog is worth more than the dog itself. Or is the protocol misaligned with economic incentives?

Update: James Kilroe’s “applications being the better investment” is valid (in my opinion) only if every dapp comes with its own ICOs and its own blockchain-agnostic token and you have a chance of encountering a killer-dapp amongst a bunch of them. Otherwise, Ethereum needs a bunch of Ethereum-exclusive dapps (thus not blockchain-agnostic) that accept only ETH for use, which I believe what Consensys is doing (correct me if I am wrong). Here, tokens would serve a totally different purpose. Today, we see most issuance of tokens from ICOs is for the purpose of fund raising. The tokens themselves are not needed at all for functions of the dapps. By right, the main function of issuing tokens is for keeping track of who owns which tokenized assets in a fractional ownership system, which I believe is where we are heading to, and much less or not at all about fund raising. In this world, Ethereum would receive full value capture. Additionally, if Ethereum blockchain can successfully resolve its trilemma, then most if not all blockchain-agnostic dapps would indirectly become Ethereum-exclusive dapps, in my opinion.

Yes, I agree that ETH price/market cap is directly related to its security, but I would love to know Vitalik’s feedback on this for confirmation or refutation. I commented about this in POS @ Low USD value of ETH? The price of ETH cannot be limited to the cost of paying for gas alone.

Same opinion here :slight_smile: The trilemma is the most priority issue in blockchain.

Our team is making EVM-compatible Gas-Delegated model which is called stamina. The problem you mentioned could be solved at least to end users and this model would lead more usages.

This is correct. There are some implecations in ETH prices such as network value, security, or the level of economic incentive / disincentive, etc. But many people don’t try to understand the meant of ETH in Ethereum deeply, and they actually don’t need to :frowning:

A lot of projects want to develop a fee-less protocol to attract attention and adoption but I really don’t think such model can be sustainable at all.

Two reasons out of many that I can think of:

  1. A decentralized system means you need to maintain a server/computer whereby it records ALL transactions done worldwide 24/7 for the rest of its existence. So if scaling becomes 1 million TPS, it means your server/computer will save up 86 billion transactions worth of memory space every day. If you are not being financially compensated for all the memory space you expend for storing transactions done by strangers around the world because you are using a fee-less protocol, then your server/computer will break down followed by the network very soon.
  2. A truly fee-less protocol, or a protocol whereby I can develop a dapp that does not accept ETH but rather accepts my own token as fee, in which that means I can create a huge amount of my own token, pump its price up sky high with hype and manipulation, and sell my token to the next sucker for real money and end up discrediting the entire system.

I am not aware of any rules in place that can prevent the 2 above from happening, but I don’t think projects hyping feeless protocol can be real or practical in real life. Which is good to say that Ethereum network needs to have fee and the fee needs to be its native coin (not some 3rd party tokens). Anyone promoting feeless protocol is probably selling a lie or a dream.

I agree the “zero-fee” fad is stupid, though the economic reasons why are a little more subtle:


One other data point regarding protocol value capture is that plain old utility token value, or in other words value through expected future use for paying transaction fees, can be quite substantial.

The total transaction fees that have been paid on the Ethereum blockchain so far can be calculated by dot-producting the raw data for these two charts here ; as of today, the result is $199.2 million, obviously with the vast majority happening within the last 12 months.


@jimmygatz I agree with most of the points that you made. I was actually thinking about writing something similar a few months ago. Especially like your visualizations of demand curves for gas, which I hadn’t thought of.

One smaller point: you can’t use discounted cash flow to value Ethereum via POS. DCF models use cash as the return, not the given asset. You hold ETH and you get paid out ETH. not in cash, so you have to create a new model which introduces certain assumptions to the value of the paid ETH (trailing 60-days price, etc.) as converted to USD.

One larger point: I think your argument on velocity can elaborate on how POS can enlist market forces to stabilize the price of ETH along the expectations for transaction fee growth

  1. Price of ETH goes down. Useful transactions on Ethereum get cheaper. Demand for ETH transactions goes up. Greater Return on ETH for staking (more transaction fees). More people stake, and more people seek to buy into ETH return. Price goes back up.
  2. Price of ETH goes up. Useful transactions on Ethereum get more expensive. Demand for ETH transactions goes down. Less Return on ETH for staking (fewer transaction fees). Fewer people stake, and more people sell out of their stakes of ETH. Price goes back down.

This feedback loop works because as ETH gets adoption, the transaction fees paid in ETH increase in inherent economic value. After all, people are choosing to buy ETH instead of keeping in fiat or some other currency for the right to acquire goods-&-services on the network. As velocity increases, these fees accumulate to those who stake, increasing the value of each stake, but also making each transaction more expensive which fewer people are willing to pay. The reverse happens as velocity decreases.

Ultimate result here is that the ETH price should on average reflect the expected earning potential on current and future transaction fees for staking minus the opportunity cost (the discount rate, essentially). Total fee $ value should roughly approximate the total economic value of current use cases on the network.

1 Like

This is essentially the “dollars have value because you pay taxes with them” argument, which I never found really convincing. Transaction fees are variable and can easily be denominated in any currency — 1 ETH does not guarantee you any fixed amount of gas, so the fact that ETH is used to pay gas shouldn’t really affect its value. Users can always convert whatever currency they use to ETH to pay gas and not care about whatever price ETH is.

As an aside, I wonder whether this whole argument is putting the cart before the horse. There’s no need for Ethereum to be “fat” in such a way that all the value is captured in the price of ethers. In fact, making Ethereum too fat probably conflicts with the idea that Ethereum has “no features”. TCP/IP, or rather ISPs don’t capture the vast majority of the value of the Internet, but that’s probably a symptom of a good thing for a foundational technology.

In general I think bottom-layer protocols attempting to capture value and interface directly with top-layer applications usually results in overcomplicated, difficult-to-adapt designs, like traditional telecom networks vertically integrating phones, phone calls, routing, etc. This is especially the case for blockchains since blockchain applications will change all the time, and trying to “stay fat” by hard-forking all the time, adding opcodes, etc to capture value actually compromises the very purpose of an immutable blockchain.

1 Like

I think, if the Ethereum network is built to be the superset of the next economic paradigm, then it is valid that Ethereum and its native coin to remain as fat as possible. The same is/was with USD and crude oil. Anyone arguing in favor of a fee-less structure with close to zero ETH value is probably the one that puts the cart before the horse. Otherwise, BTC, EOS, XRP, XLM, and many other native coins of their own blockchain should be close to zero too. You may say the same can’t be said of BTC but many others would beg to differ with their own point of argument and reasoning. I don’t think ETH should be considered a sort of 3rd-party coin, but rather a sort of “universal” coin with a very fat value that everyone has full liberty to acquire by creating something of value on top of its blockchain, in which it has no discrimination and cannot be gamed.

The USD has value not only because it is used to pay tax, but also because of all the successful businesses (and their associated value creation) on top of its USD-based national economic policy. The dollar value may be artificial now and may change in the future, but the main point is the value of a currency depends more than just being accepted to pay tax.

“Transaction fees are variable and can easily be denominated in any currency” is just the same as someone that travels to a foreign country and refuse to pay in local currency because he thinks since there is forex in place, any currency should be acceptable to the local merchant.

If a system allows more than 1 currency to be its medium of exchange, then it is effectively running a barter system, and thus inherits all the inefficiencies that come with a bartering system. Jeremy Rubin’s argument is technically in favor of a bartering system in which if I want to buy a car I must own some BMW token (example) for a purchase. And if I want to buy a phone I must own some Samsung token (example). And I cannot pay BMW with Samsung token directly (having a market to facilitate 3rd-party token exchange is not a good solution). Eventually, everyone would ask for a single medium of exchange for all of BMWs and Samsungs, in which ETH is already well in place to be one.

Note: Yes, BMW and Samsung may not necessarily create their own tokens and make it mandatorily needed for purchases, but in a system whereby anyone can create his own token for trading, BMW and Samsung are incentivized to do so. Instead of going forward, we would actually be going backward with such approach. Thus tokens should not have the function of being a medium of exchange.

1 Like

My point is that currency value is really just nominal and doesn’t necessarily correlate with how useful the currency is. ETH might really become the reserve currency of “web 3.0”, but that doesn’t really mean ETH will go to the moon since nothing requires you to hold x amounts of ethers. USD — in the Bretton Woods system let’s say to remove the factor of governments printing money to cause inflation — doesn’t go to the moon because of the extraordinary growth of the US economy. Buying index funds is way better than buying dollars as an investment in the growth of the US.

I perhaps have misunderstood what “fat” and “thin” meant. I more meant that ETH is not and shouldn’t be “fat” in the sense that buying ETH is the best way to earn a return from the growth of dapps rather than, say, investing in dapp tokens / stocks / lending money to dapp developers. It’s hard to imagine a world where all the value of dapps gets hoovered up in ETH without ETH becoming a sort of leviathan introducing new layer 1 features to preempt layer 2 innovations.

Buying index fund is definitely better than buying the inflationary dollar, but do not ignore the fact that corporations are tirelessly creating values to earn dollars.

I believe you can earn a return by 1) buying ETH, 2) investing in securities tokens (not utility tokens), and 3) creating useful dapp that earns ETH.

ETH at layer 1 is still the same ETH at layer 2. What matters is how productive are you in creating values to the society in earning your ETH.

1 Like

Yes, but there could be innovative dApps that are very ETH-frugal. For example, say somebody comes up with a new flashy STARK that is hard to efficiently implement in existing EVM. Then a dApp comes up that uses this STARK by making a metacoin-style system on top of Ethereum, where the STARKs are not computed inside the smart contract at all even though Ethereum is used like a communication bus to ensure consensus. Perhaps a sort of Truebit-style challenge mechanism is used to assure thin client security, but the high-value dApp does not use much ETH at all overall.

For a more extreme example, consider a dapp like Conifer (disclaimer: I’m the first author of that paper), which is a naming system using vaguely ENS-like semantics, with consistency enforced by blockchain rules but literally everything else enforced off-chain, and only generates one small transaction per block height. A Conifer-based naming system can scale to millions of names while not increasing its use of gas at all, and be funded using some non-ether currency.

“Thin is okay” would mean such dApps are absolutely unproblematic, while “Ethereum must be fat” would suggest hardforking Ethereum to add this new STARK as a precompiled opcode to encourage the dApp to run more directly on Ethereum and use more ETH, and, say, cripple thin clients to make Conifer-style dApps difficult to deploy.

I am truly unaware of any dapp that can run on top of Ethereum without the need for ETH in whatsoever way. If this is really possible, then the protocol is probably economically misaligned.

Obviously ETH would still be used, but it might not be used in proportion to the popularity of the dapp. Imagine a dapp using log(n) gas for n transactions processed

Thanks so much for all the responses, really fantastic discussion here!

This is effectively valuing ETH as a traditional cashflow generating asset where transaction fees are equivalent to earnings in traditional companies and a similar P/E ratio could be applied to arrive at an ETH market cap. Interestingly, this would give ETH a P/E ratio of ~90 right now which, while still high compared to most public tech companies, is not inconceivable, especially for a 4 year old high-growth startup.

While this is an interesting valuation methodology to use, I feel it doesn’t really capture the network effects (i.e. “moat” in traditional stock market speak) inherent in Ethereum as a network, both from users/developers and from security coming from scale. It also poses a problem when addressing arguments made by Pfeffer in “An Institutional Investors Take on Cryptoassets”, namely the velocity argument (ETH is working capital and agents will seek to minimise working capital) and the argument that if a) ETH’s value is derived from transaction fees paid to miners and b) mining is a perfectly competitive market such that transaction costs will trend towards the MC of mining and c) forking is a close to 0 cost activity, then d) ETH has no ability to extract economic rent and its value will simply be equivalent to the total cost of computing transactions on the network.

For me, the reason these arguments don’t make sense to me intuitively is because forking is obviously not a 0 cost activity and the cost of forking is basically equivalent to ETH’s network effects which in turn determines the economic rent ETH is able to charge. However, these dynamics are not present in traditional joint stock corporations and I’m unsure how they can be adequately modelled using the P/E ratio mentioned above. Are these network effects simply reflected in a higher P/E ratio? Has anyone done any interesting work in terms of coming up with an appropriate transaction fee multiple to apply?

Thanks for this, great response! Is the main reason a DCF couldn’t be used the fact that when people receive their quarterly block rewards, they may all rush to sell simultaneously, thus crashing the price? If so, wouldn’t this eventually be built into market expectations such that this would no longer happen?

As for your larger point, I totally agree. PoS gives ETH a yield and makes it a cashflow-generating asset, meaning there’s a natural price floor which is the price at which the yield from staking becomes greater than some X (whatever the risk-adjusted market discount rate is for ETH). At this point, not only will arbitrageurs buy in to pocket the extra yield, but as you mention there will also be the positive feedback loop of transactions being cheaper and thus increased demand for them. Similarly, there should theoretically also be a price ceiling at which point expected risk-adjusted yield will become so low that stakers will sell their stakes in addition to the positive feedback loop of transactions becoming more expensive and demand decreasing. Interesting stuff!

I will note that the transaction fee argument only works if ETH moves to PoS. If it goes to miners, then yes, I agree it depends on complicated spooky stuff about velocity.