Originally published: November 2021
I’ve been asked a number of times for an update for my views on Ethereum since my January 2021 article on the topic, which received over a quarter million reads.
In that prior article, I described Ethereum, explained areas where I was bullish, but also expressed my fundamental concerns with it. The overall tenor of the article was somewhat critical of Ethereum, which is why it received so much attention. In that article, I was also quite bullish on the proliferation of stablecoins in the coming years.
I don’t mean to pitch my content here, but for context, members of my premium service already know my updated views on Ethereum, because I’ve been providing updates on Ethereum pretty much every month since that initial article.
The summary of those many reports was that I frequently described problems with the Ethereum blockchain including DeFi hacks, centralization issues, unintended chain splits, NFT speculation and so forth, but that I have been rather tactically bullish in terms of price action for the intermediate term, once I initiated regular coverage in January. Here are some brief excerpts:
For those that are watching it, a firm Ethereum price break over $1,400 should be pretty bullish for the protocol in the intermediate term, since it clears out overhead resistance.
-January 31, 2021
Statistically speaking, ethereum and other alts could very well outperform bitcoin to the upside during the bull period of the cycle as they often do, but I’d be worried about a lot of digital assets, especially outside of bitcoin itself, on a down leg of the cycle in maybe 2022 or 2023.
-February 14, 2021
The amount of ETH on exchanges has been in a downtrend since August 2020, similar to what’s happening with BTC. All else being equal, that’s bullish.
-April 14, 2021
While I have concerns about Ethereum’s long-term design and shift towards Ethereum 2.0 (the very ability to change its monetary policy shows how impermanent the monetary policy is), it’s hard to be bearish on price action in the intermediate term. EIP 1559, which I wrote about favorably in my otherwise somewhat critical public piece on Ethereum back in January, should be pretty bullish for price when it goes into effect. And with Ethereum 2.0 staking in effect since December 2020, ETH tokens continue to leave exchanges and get locked up. Implementing EIP 1559 while delaying the shift towards Ethereum 2.0 (which is basically the plan at the moment) is actually kind of a perfect storm for price in a positive way for the year ahead. So, while I have higher conviction on BTC than ETH with say a 5-year view, the specific dynamics holding up ETH prices are pretty strong for the back half of this year. It would need to break above $2,900 to get interesting again, though. Right now it’s in a sideways consolidation.
-June 6, 2021
Overall, Ethereum still remains under a persistent supply squeeze at the moment (one-way staking until Ethereum 2.0 is launched, with about 7.4 million locked-up ETH at the moment), so I remain somewhat tactically bullish on price despite being long-term cautious about some of the technical fundamentals, use-cases, competitors, etc.
-September 5, 2021
I remain tactically bullish on ether in terms of price action despite some misgivings about the long-term risks and use-cases. My tactical bullishness is in part based on the Ethereum 2.0 lockup contract that continues to soak up a ton of ether (up to over 8 million now, which can’t come out), and as a consequence of that, ether continues to quickly get drawn down from exchanges even a bit faster than bitcoins are getting drawn down from exchanges. It’s a well-engineered supply squeeze.
-October 31, 2021
And it seems my view on stablecoins from the original January 2021 article was accurate, since they increased from $33 billion at the time of writing to $140 billion in capitalization in under a year:
Stablecoins are particularly important, in my view. I’m bullish on the amount of money locked up in stablecoins. It’s a space to watch, for both good developments and bad developments. The US Office of the Comptroller of the Currency now officially permits US banks to use stablecoins. They’re a much more liquid form of fiat currency, and can have various implications for central bank digital currencies and the existing global monetary system.
-January 17, 2021
Chart Source: The Block
There has been a bit more institutional interest in Ethereum, as well as institutional interest in follow-up chains like Solana, than I expected from January, so that’s something I’ve been monitoring. There is still a lot of regulatory uncertainty around these types of tokens; unlike Bitcoin they generally seem to meet the definition of being financial securities.
For this follow-up public article, I figured it’s time to delve into three related concepts that are more broad than just Ethereum. The first is about the trade-offs of proof-of-stake as a consensus mechanism in general, the second is the stablecoin centralization problem, and the third is the spectrum of centralization that various smart contract chains use to compete with each other on fees.
All three of these relate together because they affect how truly decentralized a proof-of-stake smart contract blockchain can be compared to the Bitcoin network, and how they can perform relative to each other in hostile or non-hostile regulatory environments.
So, this article can help articulate some of the long-term concerns I have with the various smart contract blockchains, even at times when I am bullish on price from a tactical sense, and in the context of the fact that I am interested in the concept of smart contracts generally.
I want to re-iterate that I try to be as objective as possible when analyzing blockchains. It’s no secret at this point that I like the Bitcoin protocol quite a bit, but that’s because it’s the one I am able to find the fewest faults with on a risk-adjusted basis. I analyze multiple asset classes, from stocks to bonds to commodities to digital assets, and often compare individual assets within those asset classes. So when I analyze blockchains, I approach it the same way.
And importantly, I separate technical price action from fundamental soundness, because they can be very different things for periods of time. This article touches on blockchains like Ethereum and Solana, but more broadly is about the centralization problems of proof-of-stake blockchains and custodial stablecoins in general, as the topics discussed here can apply to many blockchains outside of Bitcoin.
- Proof-of-Stake vs Proof-of-Work
- The Stablecoin Centralization Problem
- How Important is Decentralization?
- Protocol or Operating System?
Proof-of-Stake vs Proof-of-Work
Satoshi Nakamoto’s Bitcoin protocol comes to a consensus on valid transactions using a method known as proof-of-work. Satoshi Nakamoto referenced Adam Back for his prior development of proof-of-work as one of eight references in the Bitcoin white paper.
Since then, a number of people have proposed that other methods of consensus, such as proof-of-stake, are more efficient. Often the advantages of these proposals are described with insufficient acknowledgement of the trade-offs that they make compared to proof-of-work. This section explores that concept.
(The summary of this chapter, for those that want to skim parts of this otherwise long article, is that proof-of-work is inherently money-like while proof-of-stake is inherently equity-like.)
The Bitcoin network is programmed to create a new block on average every ten minutes and add that block to the blockchain, which consists of hundreds of thousands of blocks since inception in 2009.
A new block is produced by a bitcoin miner (a specialized computer) contributing processing power (and thus electricity) to solve a cryptographic puzzle that the previous block created, at which point the miner can package thousands of bitcoin transactions currently in the queue, into that block. That’s how transactions get settled. The network is programmed to target average block times of ten minutes, meaning on average every ten minutes a block of thousands of transactions is added to the blockchain.
Processors use random guesses to solve the puzzle left by the prior block, but the law of large numbers is such that the more bitcoin mining equipment you have, the more blocks you find over a sufficiently long period of time.
If miners drop off the network and new blocks on average start taking longer than ten minutes to produce, the network is automatically programmed to make the puzzle easier by a quantified amount, so that blocks go back to an every-ten-minute average schedule. Likewise, if a lot of miners join the network and blocks get added to the blockchain faster than every ten minutes on average, the network will make the puzzle harder. This is known as the “difficulty adjustment” which occurs automatically every two weeks, and is one of the key programming challenges that Satoshi Nakamoto solved to make the network work properly.
So, at any given time, there are millions of bitcoin mining machines around the world looking to solve the puzzle and create the next block, and there’s a natural feedback mechanism to ensure that blocks are created on average every ten minutes, regardless of how many or few miners are on the network.
In the first half of 2021 China (by far the largest country in terms of miner concentration at the time) banned crypto mining, and approximately half the global Bitcoin network went offline and started moving elsewhere. Bitcoin’s payment network briefly slowed down, but otherwise kept working with 100% uptime. The difficulty adjustment then kicked in, and brought the network back up to its target speed. Imagine if Amazon or Microsoft were told with one week’s notice that they had to move half of their server capacity internationally; they would likely experience uptime issues for their services for the rest of the year (at least) as they moved and rebuilt half of their infrastructure. The Bitcoin network instead continued to operate with 100% uptime.
If a miner creates an invalid block, meaning one that doesn’t conform to the shared rules of the existing node network, the network discards it. If two miners produce a valid block at around the same time, the winner will be decided by which one gets found by the rest of the network first and has another valid block produced and added onto it, becoming the longer (and thus official) blockchain. If those second blocks are also close, then it will come down to who wins the third valid block, or fourth valid block. Eventually a longer chain wins, as a greater share of the network is finding it and building on top of it.
This process is known as “proof-of-work”. Millions of machines are using electricity to apply processing power to guess the answer to cryptographic puzzles left by the most recent block. This may seem like a waste of energy, but it’s what keeps the system decentralized. Work is the arbiter of truth, in this case. There is no central authority that decides what constitutes a valid block or a valid set of transactions; the longest blockchain is verifiable at any given time, and is recognized as truth by the rest of the network based on code.
The longest blockchain is the one with the most work put into it, and that also meets the consensus criteria that the node network checks. That blockchain becomes the global consensus.
The more energy that Bitcoin’s network uses, the more secure that its latest transactions are against most types of attacks. Many of the tiny non-Bitcoin blockchains have been victims of 51% attacks, where a single entity temporarily or permanently gains control of over 51% of the processing power on the network, and uses that majority of processing power to re-organize blocks and perform double-spend transactions (which is essentially theft).
This chart, for example, shows Bitcoin’s network processing power compared to the processing power of some of its hard fork copycats:
Chart Source: BitInfoCharts.com
Both of those other blockchains only have 1% or less of the Bitcoin network’s total processing power, and have been hit by malicious block re-orgs. In fact, if just 1% of bitcoin miners decide to do a 51% attack on either of those two hard forks, they can. The same is not true for the other direction, since it is the Bitcoin network that has a far larger network of miners and energy usage than them, by two orders of magnitude.
That shows the importance of network effects in the blockchain industry, and why Bitcoin’s energy usage has kept it uniquely secure.
When someone asks, “can’t you just copy Bitcoin?”, that’s why the answer is “no”. You can replicate the open source code, but you can’t replicate the fact that millions of ASIC miners are securing the Bitcoin network and not your copycat network, you can’t replicate the fact that tens of thousands of full nodes are ensuring consensus, and you can’t replicate the fact that thousands of developers are working on making the Bitcoin network better every day rather than working on your copycat network. And Bitcoin’s second layer, Lightning, has a lot of open channels and liquidity that can’t be easily replicated either; it took years to build.
Trying to copy Bitcoin would be like if I copied the content from Wikipedia and hosted it on my website. Technically it could be done, but it wouldn’t do much. It wouldn’t gain the real Wikipedia’s traffic, because it wouldn’t have the hundreds of millions of links pointing to it from other websites. And it wouldn’t be updated like the real Wikipedia, because there’s no way I could convince the majority of those volunteer editors to come work on my version instead. Unless I could somehow succeed in the herculean task of convincing the majority of the network to move over to my version, it would always just be a shadow of the real one with a tiny fraction of the value.
The same is true if I made a poor mimic of Twitter. I could make it look like Twitter, but it wouldn’t really be Twitter, full of users and developers.
Okay so as we discussed, proof-of-work is a system where miners compete with electricity and processing power to build the longest blockchain, which becomes the accepted blockchain. The digital blockchain, via proof-of-work, is thus connected to real-world natural resources.
The Bitcoin network has operated via proof-of-work since inception in 2009, and with no plans to change that.
The Ethereum network has also operated via proof-of-work since its inception in 2015, but has been planning to change to a proof-of-stake system for several years.
Many newer smart contract blockchains launched after Ethereum have incorporated proof-of-stake consensus from the start, which puts them ahead of Ethereum in that regard, but without Ethereum’s substantial network effect.
So, let’s dive into how proof-of-stake works.
Proof-of-stake is a system where holders of the cryptocurrency lock up or “stake” their coins, and use them to vote on the valid blockchain, and get rewarded with more coins for successfully creating new blocks. Instead of committing electricity and processing power to create new blocks on the blockchain, they’re committing their stake of coins to do so.
Proof-of-work is simple, because there is no need to punish bad miners that try to validate the wrong chain or make invalid blocks that don’t fit the rules of the node network. Their punishment is simply that they spent electricity on blocks that weren’t valid or weren’t included in the longest eventual chain, and thus lost money. They self-inflict their own wound, and thus it rarely happens on purpose. There is a tangible connection between the blockchain and real-world resources.
Proof-of-stake is more complex, because there is no connection to real-world resources and the system needs a way to punish stakers that improperly vote on the “wrong” chain. In addition, they need a way to make sure stakers aren’t voting on all possible chains (which can’t be done with proof-of-work, because it takes real-world resources for each one). So, proof-of-stake is a much more complex system that will try to take away stakers’ coins if they vote improperly, and has ways of checking to see if they are voting on multiple chains.
Ben Edgington, a developer for Ethereum and someone who is in favor of Ethereum’s upcoming shift towards proof-of-stake, went on the Compass Mining Podcast and elegantly explained the long-term challenges that Ethereum has faced as it undergoes its multi-year (and long-delayed) shift from proof-of-work towards proof-of-stake:
The reason it has taken a while, you know we’ve relied on proof-of-work in Ethereum for five plus years, is that proof-of-stake is complicated. Proof-of-work is fundamentally very simple, is easy to analyze, is easy to implement and deploy, and proof-of-stake has a lot of moving parts. You can code up a proof-of-work algorithm in a hundred lines [of code] or so. Our current clients are a hundred thousand lines or so for proof-of-stake.
And I think the theoretical foundations for proof-of-stake have taken time to mature. It’s not obvious how to make it robust, there are attacks like long-range attacks and things that just don’t exist in proof-of-work, that we’ve had to think through and come up with solutions to, so that’s just taken time. So we’ve relied on the tried and tested proof-of-work algorithm and it served Ethereum well.
The host in that podcast discussed how early proponents of the Bitcoin network were initially interested in proof-of-stake but determined it had too many attack vectors. He then asked Ben how Ethereum and the proof-of-stake model defends against those attack vectors. Ben thinks it is robust and is in favor of it, and described proof-of-stake’s workarounds as follows:
The initial difficult one to solve was what we call “equivocation” which means that it is basically costless to produce blocks, so if I am a proposer of a block, I can propose two competing blocks, or three, or a hundred, and broadcast them to the network that has no real way to distinguish between these blocks. That can be extremely disruptive and attack the chain, certainly split the chain, and so, we deal with this through a mechanism we call “slashing”. And so for a proposer to propose conflicting blocks, that is a slashable offence. The network can detect that. Another proposer can come along and say here are two blocks that were proposed by the same validator at the same time, their signature is on it, so that can’t be faked, so here is a proof that they acted incorrectly. And then part of their stake is taken away from them and they are rejected from the network at that point. So you only get once chance. In proof-of-work, if your 51% attack fails, you can just crank it up and do it again and again. In proof-of-stake, you get one chance, you’re slashed, you’re out of the network, and your ETH is locked up for a while, and so it’s kind of self-healing, in that respect. So that was a major theoretical breakthrough that kind of made people think, “actually we can kind of do this, there are fixes to common attacks.”
Another one is called a long-range attack and this is kind of subtle, but the idea is that once you’ve exited the network as a validator, you can then go back in time, effectively. So I exit the network and I can go back a month in time, and produce (if I have enough validator keys) as many historical blocks as I wish, I can write a different history for the chain, effectively, which conflicts with its current history, and I’ve exited so I can’t be slashed anymore. So that’s a long-range attack. We have an analysis of this, and an understanding of this, which Bitcoiners will hate but we call it “weak subjectivity”. It’s the idea that anybody who is continually online is always safe, because they are monitoring the chain and they always know what the correct chain is. If you sync from scratch, you know you sync from genesis, there is a danger that you follow an attacker chain, so you need a checkpoint, which guarantees that you are on the right chain, which you need to get from someone who has been online for the entire period or somebody who is guaranteed to be on the right chain. Now that is called “weak subjectivity”. There are rules about how frequently these checkpoints need to be produced, how we can rely on them, and we are building “somewhat trustless” mechanisms for getting hold of these checkpoints. It is, I understand, a deep clash with Bitcoin ideology in that sense that anybody in vacuo should be able to sync up from genesis and know they are on the right chain without trusting anybody in any way, shape, or form. We’re not doing that. That seems to be very difficult with proof-of-stake, that’s a compromise we made, but we believe in practice this is completely workable and will not lead to any practical attacks of any sort.
Besides this larger amount of complexity, trust, and attack surfaces, I would argue that a main issue with proof-of-stake is that it can be prone to centralization.
With a proof-of-stake system, the more coins you have, the more voting power you have, and those with the coins are also the ones earning the new coins from staking. Since they don’t need to expend resources to stake, they can simply increase their overall staking amount as they earn ongoing coins from staking rewards, and exponentially grow their influence on the network over time, forever. Network dominance tends to lead to more network dominance, in other words.
It would be like a political system where you get a vote for every hundred dollars you have, and then also get paid a dollar by the government for casting each vote. Mary the high school science teacher with $20,000 in net worth gets 200 votes, and earns $200 from the government for voting. Jeff Bezos, with $200 billion in net worth, gets 2 billion votes, and earns $2 billion from the government for voting. He’s a more valuable citizen than Mary, by a factor of a million, and also gets paid more by the government for already being wealthy.
That’s not a system many folks would like to live in. Eventually it would likely consolidate into an oligopoly (if it wasn’t already), with a handful of multi-billionaires controlling most of the votes and ruling everything. If it gets too centralized, that kind of defeats the purpose of a decentralized blockchain.
Instead, that proof-of-stake system mainly works well for stakes in centralized private property, like corporations. In a corporation, each share is worth a vote for proposals and board seats, since the owners decide what the company will do in proportion to their ownership. These are voluntary organizations; shareholders, customers, and employees can go to a different corporation if they don’t like the rules. That’s different than a national election, which is supposed to be a decentralized platform. And it’s different than money or legal tender as well.
So, I don’t consider the proof-of-stake model bad for other cryptocurrencies to use in terms of experimentation, if they are more like a corporation. In fact, proof-of-stake can increase the cost of attacking the protocol, since an attack or group of attackers would need to acquire a lot of the coins (unless they find and exploit a bug due to the greater attack surface, or somehow steal the coins). There are certain Defi projects or platforms, for example, that can operate like a company and use proof-of-stake to be efficient and costly to attack, if all goes well. They’ll be prone to centralization, but if they’re voluntary services competing with other proof-of-stake services, that can be okay. If their service isn’t good, people can go elsewhere. In general, we have no problems with companies being centralized, because they are companies.
Instead, proof-of-stake mainly seems less suitable for a decentralized and censorship-resistant global monetary asset, especially when considered along with the issues that I’ll describe in the second half of this article about stablecoins. Proof-of-stake is inherently equity-like rather than money-like, compared to proof-of-work.
Adam Back described this succinctly a while ago:
You see that with other commodity money, like physical gold. It’s a system that works because money has a cost. I think money that doesn’t have a cost ultimately ends up being political in nature. So people closer to the money, the so-called Cantillon Effect, are going to be advantaged.
How Bitcoin Survived By Not Being Proof-of-Stake
In a proof-of-work system and particularly the Bitcoin network with its purposely-small nodes, power is distributed between miners, developers, and individual nodes.
Your ability to be a miner is based on your ability to put forth capital and find low-cost electricity. Rather than the entrenched miners having an advantage and increasing their advantage over time (as is inherently the case with proof-of-stake systems), newer miners actually have a technical advantage over existing miners in some ways because they buy the newer machines with more processing power per watt, thanks to Moore’s law, with no existing sunk cost. Mining businesses, old and new, are all constantly refreshing themselves with capital expenditures, making use of new cheap or stranded energy resources. Management quality and experience is critical, and economies of scale only get you so far.
Plus, the Bitcoin network’s designers went to great lengths to make it easy and cheap to run a full node (unlike almost any other cryptocurrency), which allows any user to audit the entire blockchain and reject blocks that don’t conform with the rules of the node network. In the Bitcoin network, the real power rests with the nodes, rather than the miners. If miners try to collude and mine blocks that are invalid, the node network simply rejects those blocks.
We can think of this as being similar to the US Constitution that set up three branches of government to limit each other. The Executive Branch, Legislative Branch, and Judicial Branch have various ways to overrule each other in certain contexts, and have staggered term limits, which by design makes the political system resistant to changing too quickly and thus devolving into either authoritarianism on one hand or mob rule on the other.
Similarly, the Bitcoin network has the node network, the miners, and the developers, with the node network being the final arbiter of consensus but relying on miners to order transactions and developers to create updates that both the miners and node network accept as being improvements. The natural state of the network is to resist change, especially changes to the foundational design of the system, so it requires overwhelming consensus to change something, and those changes are backwards-compatible soft forks that nodes can choose to opt into or not while still being compatible with the protocol.
Many other blockchains that have come into existence since the Bitcoin network make multiple trade-offs including making the nodes require immense processing power, bandwidth, and storage, so that only industrial-scale entities can run them, which centralizes the network into a handful of major providers who can audit the blockchain and ensure consensus.
Bitcoin’s proof-of-work and small block design keeps a lot of power with the individual users. Anyone running a full node can audit the entire blockchain, verify their individual transactions, and participate in the network effect that ensures consensus.
I recommend that folks interested in Bitcoin and the broader cryptocurrency space read The Blocksize War, which is a 2021 book that chronicles the history of the Bitcoin network as different factions struggled with each other to shape the design of the protocol, and to see who had the power (developers, corporate miners/exchanges, or individual users/nodes). It was a real-world test of Bitcoin’s level of decentralization. It was a constitutional crisis for the Bitcoin network in other words, and it passed the test.
Ever since the network’s early history, there was a growing divide between people who wanted to increase the block size and people who wanted to keep it small. Increasing the block size allows the network to process more transactions per unit of time (not taking into account layer two solutions and side-chain solutions Lightning and Liquid, which didn’t exist yet). However, increasing the block size also increases the bandwidth and data storage required to run a full node, and thus puts it out of the reach of the everyday user on a laptop or Raspberry Pi.
Even Satoshi Nakamoto himself played a dual role in this debate; he’s the one that personally added the block size limit after the network was already running, but also discussed how it could potentially be increased over time as global bandwidth improves.
If users can neither mine nor operate a full node themselves, they have to trust large-scale network providers, and Bitcoin would cease to be a trustless, decentralized system. It would permanently weaken the consensus function of the node network, in other words.
After the seeds of this disagreement were laid from the protocol’s inception, and with Satoshi Nakamoto long gone, it was from 2015 through 2017 that the blocksize war went into full conflict.
At one point in 2017, over 80% of miner processing power, the biggest maker of bitcoin mining equipment, prior lead developers of Bitcoin, and a large number of major custodians and exchanges including Coinbase and Grayscale, were in favor of increasing the block size with an upgrade called SegWit2x (not to be confused with the normal SegWit update). That’s an overwhelming amount of support among the corporate-level players in the industry, or as they described themselves in their New York Agreement, they were “a critical mass of the bitcoin ecosystem”.
And yet, they failed.
The existing developers and most importantly the majority of the individual node operators were not on board with the plan, and so along with multiple other reasons, it was aborted.
SegWit2x, (abbreviated B2X or S2X, and originally called SegWit2Mb), was a failed contentious hardfork attempt outlined in the New York Agreement that intended to double the block size limit. The hardfork has been denounced as an attempt made by CEOs and owners of large Bitcoin businesses to introduce changes to the currency’s protocol and development cycle with ulterior motives.
Though over 80% of miners signaled intention for SegWit2x and the New York Agreement, it failed to gain any consensus among the community and Core developers.
If Bitcoin had been proof-of-stake, and without the real power in the Bitcoin network being among the individual node operators (with those nodes specifically being designed so that anyone can run them), those big corporate players might have been successful at reshaping the Bitcoin network. That could have put running a full node out of reach of normal users and thus partially centralized the protocol. More realistically, it may not have been that small block size increase that did it, but it could have set the stage for a series of much larger block size increases down the road.
If Bitcoin were built on a proof-of-stake model, where the more coins you have the more votes you have on how the network functions, the large exchanges and custodians could have used the millions of coins they held on behalf of clients to vote in their own favor. This is similar to how Vanguard and BlackRock hold trillions of dollars of indexed equity assets for their users and maintain the voting rights on those assets.
Some folks on the big block side also forked their own blockchains out of Bitcoin throughout this war, creating large-block versions of Bitcoin, including Bitcoin XT, Bitcoin Classic, Bitcoin Unlimited, Bitcoin Cash, and Bitcoin Satoshi Vision. All of those have fallen significantly vs Bitcoin in terms of market capitalization and hash rate, as they have been rejected by the market. Some of those versions are now dead, and others have been subject to major 51% attacks.
Proof-of-work and small full nodes together are the main way currently known to keep a blockchain sufficiently decentralized on the base layer, and with the highest level of security including the most hardened attack surface. If Bitcoin ever upgrades to a different system, it would only be with overwhelming consensus among users.
Proof-of-Stake Technical Challenges
Ethereum has been running into more acute scaling problems than Bitcoin, which has opened the door to a number of more centralized (and thus in some ways more efficient) competing smart contract blockchains.
And among those new proof-of-stake competitors, there are multiple examples of their systems running into technical problems.
One of the highest-profile issues involved the Solana blockchain going down for 17 hours, requiring a manual coordinated restart by validators. Solana is a popular VC-backed smart contract blockchain that tries to be a lot more scalable than Ethereum by implementing a combination of proof-of-stake with proof-of-history to achieve significant throughput.
What are its trade-offs? Well there are a number of them. Solana’s higher throughput compared to Ethereum does not come with a free lunch.
First of all, to run a Solana validator you need a computer with 12 CPU cores, 128 gigabytes of RAM, and 300Mbit/second upload speed (1 Gbit/second recommended). That setup, especially the upload speed part, basically means you need to be a datacenter operator to run a Solana validator. Unlike Bitcoin, you can’t use a laptop at home to validate the entire blockchain. Solana is not auditable to you, in other words.
Secondly, even those datacenter-level validators have to rely on archivers if they want to go back through the full history of the blockchain, because over time the amount of stored information becomes utterly massive. Bitcoin does not have this problem; after 13 years of operation the entire Bitcoin blockchain can be stored on a common computer drive. In another 13 years, Bitcoin will still be storable on a common computer drive. Solana’s archival history after a decade or two would be an astounding amount of data that again makes it non-auditable to you.
Thirdly, Solana uses manual slashing. In other words, it’s a blockchain that requires human decisions to determine consensus if there are significant attacks on the network:
Slashing is a hard problem, and it becomes harder when the goal of the network is to have the lowest possible latency. The tradeoffs are especially apparent when optimizing for latency. For example, ideally validators should cast and propagate their votes before the memory has been synced to disk, which means that the risk of local state corruption is much higher.
Fundamentally, our goal for slashing is to slash 100% in cases where the node is maliciously trying to violate safety rules and 0% during routine operation. How we aim to achieve that is to first implement slashing proofs without any automatic slashing whatsoever.
Right now, for regular consensus, after a safety violation, the network will halt. We can analyze the data and figure out who was responsible and propose that the stake should be slashed after restart. A similar approach will be used with a optimistic conf. An optimistic conf safety violation is easily observable, but under normal circumstances, an optimistic confirmation safety violation may not halt the network. Once the violation has been observed, the validators will freeze the affected stake in the next epoch and will decide on the next upgrade if the violation requires slashing.
In the long term, transactions should be able to recover a portion of the slashing collateral if the optimistic safety violation is proven. In that scenario, each block is effectively insured by the network.
So, Solana is not really even an automated blockchain. It’s a step back from both proof-of-work Bitcoin and proof-of-work Ethereum in terms of automation, in exchange for more throughput and low fees. The consensus mechanism is more manual, human, and political.
And before the Solana bulls get mad at me, I’ll point out that I’m not biased against Solana. Smart contract platforms have a natural tendency to move towards greater and greater centralization, because the more centralized they are, the more efficient they are, and users want efficiency (e.g. low fees and fast confirmations). Back in early September 2021 when Solana was at $40 billion market capitalization and Cardano was at $90 billion market capitalization, I suggested an eventual flippening in my research service, which was two months before it happened in early November 2021:
I think Solana (currently the 7th largest crypto by market cap) has a decent shot of catching up with Cardano (currently the 3rd largest) in terms of market capitalization, although I don’t invest in either of them. The main draw for Solana right now is that they are building out semi-decentralized exchanges and other applications, and their fees are much lower than Ethereum.
-September 5, 2021
And maintained this view even after Solana broke for 17 hours:
Meanwhile, Ethereum’s competitor Solana ran into more severe issues. The entire blockchain went down for nearly a day on September 14th due to an overload in transactions. Validators had to coordinate and restart it. It’s funny timing because back in my September 5th report, I discussed how centralized Solana is, and this type of problem is an example of that. I still think Solana has a reasonable shot of catching up with Cardano’s market capitalization.
-September 19, 2021
Basically my thesis with Solana was that most users of smart contract platforms care more about low fees than high levels of decentralization, at least in a non-hostile regulatory environment. I had already been observing this with Tether stablecoins shifting from Ethereum to Tron when the fees became high on Ethereum.
As a result, smart contract platforms that have higher throughput and a critical mass of support, each have a decent shot of taking market share. The field at this time keeps diluting itself with cheaper and more centralized networks.
However, Ethereum also goes partially down from time to time with unintended chain splits, and it’s possible that if it changes over to proof of stake, it could face similar more severe issues to what Solana faces. (In contrast, Bitcoin has had literal 100% uptime since spring 2013, and when it had an issue back then, it was worth less than $1 billion in market capitalization and thus was truly in the experimental stage).
An October 2021 paper from Stanford (and financially supported by the Ethereum Foundation, to their credit) called Three Attacks on Proof-of-Stake Ethereum outlined ways to attack the system once Ethereum switches over to proof-of-stake. I’ll let the PhDs of computer science determine which attack paths are valid and which are defendable with upgrades once the attack is known, rather than go into depth on that paper. I suggest reading it.
Ethereum developers have delayed Ethereum’s transition to proof-of-stake for years (their initial plan to switch to proof-of-stake was back in 2016 and we’re about to enter 2022), acknowledging that it’s a much more complex system than proof-of-work. Solana developers, likewise acknowledge how hard slashing (a necessary component of proof-of-stake) is to implement, and have manual slashing which very much centralizes the system, along with the lack of true auditability by most participants.
Hugo Nguyen has a series of articles (such as here and here and here) critiquing proof of stake in terms of first principles. The main theme is that by not including unforgeable costliness as part of their designs, proof-of-stake systems are inherently more circular in nature and thus rely more on some degree of constant trust, have less ability to recover from chain splits without manual intervention, and have limited ability to secure the historical blockchain. An excerpt:
Second & much more importantly, once the PoW node software has been downloaded, it’s reasonably safe for the PoW node operator to turn off the node for an arbitrary amount of time. Past the bootstrapping stage, PoW is highly permission-less: nodes can come & go whenever they like. The only exception to this is in the event of hard forks, which require the node operators to repeat the bootstrapping process (another reason hard forks should be used very judiciously & avoided if possible).
In contrast, a PoS node operator, even with the correct software downloaded, will regularly need to reach out to trusted third parties to ensure he stays on the canonical chain. The fear of losing contact with the main network & getting tricked onto the wrong chain will continue for eternity, possibly long after the trusted third parties cease to exist! This marks a significant degradation in security.
A lot of people offhandedly propose proof-of-stake as being superior or better technology than proof-of-work, and praise higher-throughput systems, without realizing these technical issues at all. Many of the things they think are bugs to be eliminated from the system, like the fact that a proof-of-work system has a real-world resource cost, are actually features that make it as secure as possible.
And then they are surprised that many people don’t view tokens of proof-of-stake protocols and higher-throughput systems as being secure enough to be considered “global money” or “pristine collateral” in the same way that bitcoins are. Instead, these types of protocols are rather centralized experimental platforms for smart contracts, which can be speculated upon like tech growth stocks but ideally only by those who fully appreciate the risks.
The Stablecoin Centralization Problem
Stablecoin custodians represent another attack vector and centralization problem against smart contract platforms that have DeFi as a key part of their ecosystem, whether they are proof-of-work or proof-of-stake. This problem affects protocols like Ethereum and Solana, but not really Bitcoin.
(The summary of this chapter, for those that want to skim parts of this otherwise long article, is that any smart contract blockchain that relies heavily on DeFi for its use case, can have the outcome of its hard forks significantly determined by centralized stablecoin custodians. These custodians can nullify the value of all stablecoins on whichever side of the fork they don’t view as the correct one, which severely reduces the survivability of that side of the blockchain by rendering its DeFi mostly insolvent. This can include picking the forked chain over the original chain, and therefore all variables of the blockchain are potentially mutable even if the node network doesn’t like the changes.)
Stablecoins are tokens on a blockchain that represent units of fiat currency, and most commonly, the U.S. dollar. Now that smart contract blockchains exist, they can be used for various purposes. One popular purpose is that an entity collects dollars, and then issues tokens on a smart contract blockchain that represent redeemable claims on those dollars, and these tokens are called “stablecoins” because they are stable against that dollar, and are ostensibly backed 1-for-1 by dollars and dollar-equivalents (although that last part has historically been quite controversial, since that’s not always the case).
Once stablecoins are issued, people can then use whichever blockchain they are issued on to send and receive stablecoin payments between themselves with no centralized third party. From a user standpoint, stablecoins are a significant technological leap over existing bank payment systems, especially for international payments of any size, or large domestic payments. You can send someone a million dollars on another continent at 2am on a Sunday night and they can receive it in minutes, and you can verify the transaction on the blockchain. (And that, by the way, is part of why governments are not particularly thrilled with their existence, and are working on regulations to get them increasingly surveilled and censorable).
These types of stablecoins are of course quite centralized. The custodian holds the actual money; the collateral that backs all of those tokens. The custodians have the power to “blacklist” some their tokens, which freezes them and basically makes them worthless. Tether has blacklisted over 500 addresses and counting:
At the end of the day, the custodians determine which of their token liabilities meet their criteria to be redeemable, or even to be sent among peers. If you do something they (or their governments) don’t like, or your tokens are on the wrong side of a hard fork of what the stablecoin issuer believes the preferred side of the fork to be on, your money might not be worth anything anymore.
There is now over $140 billion in stablecoin value on smart contract networks. This gives them tremendous power over the networks. To explore why, let’s review the concept of a blockchain hard fork.
Hard Forks: Reviewed
A blockchain can have something called a “hard fork”, where developers and miners/validators decide to change the protocol rules and create a new set of blocks that don’t conform to the rules of the existing node network.
If there is a significant number of miners that agree on these new changes, they can sustain this new blockchain indefinitely. These changes could include major modifications of the money supply, block size, issuance rate, and other foundational rules of the protocol. Meanwhile, if other miners also continue to create blocks that conform to the existing node network, then the singular blockchain splits into two, like a fork in the road. The original blockchain and the new blockchain both continue in parallel.
Bitcoin Cash is a well-known example; they significantly increased the block size compared to the original Bitcoin protocol, and went in their own direction, and subsequently lost a lot of value compared to Bitcoin. Bitcoin Satoshi Vision forked out of Bitcoin Cash and subsequently lost a lot of value compared to Bitcoin as well.
The reason that Bitcoin is often called “immutable” by its proponents is that it is extremely resistant to changes. Once you have a full node, you have the software that recognizes blocks as either being valid or invalid according to the consensus rules of the protocol, such as block size, money supply, etc.
If someone makes a hard fork, they basically just make their own blockchain and it doesn’t affect yours, and doesn’t affect the software consensus that is the Bitcoin network. So far, every hard fork attempt on Bitcoin has been unable to gather the critical mass of users to move over to it.
If developers and miners on your blockchain decide to create a soft fork (a backward-compatible smaller change that does conform to the rules of the existing node network but also narrows them), then they can do that, and you can operate with the network whether or not you personally decide to upgrade to that new subset of rules that constitutes the soft fork or not. Either way, you’re still compatible with the network.
This is why, in the 2015-2017 block size wars, extremely powerful forces could not overcome the power of individual users running their own nodes. The majority of miners, the near-monopoly producer of mining equipment at that time, several of the biggest bitcoin-related exchanges and companies, and some of the influential early developers, all tried to change the Bitcoin network to their preference, and were rejected.
It’s hard to describe how big of a combined assault that was. It was like the movie Avengers: Infinity War where the entire team of Avengers including Iron Man, Thor, Hulk, Captain America, Black Widow, Black Panther, Spiderman, the Guardians of the Galaxy, Scarlet Witch, Vision, and Doctor Strange teamed up against Thanos and… still lost against Thanos.
Thanos was inevitable in that movie. Likewise, Bitcoin was immutable, thanks to its user-led node network. And it proved it in the field in 2017. It doesn’t mean it would resist every challenge but with this event, it has withstood a far bigger challenge to its decentralization model in the field than any other cryptocurrency.
Before and after failing to change the Bitcoin network, many of those people created numerous hard forks of Bitcoin, with the most well-known one being Bitcoin Cash. When a hard fork happens, each user keeps their existing coins (and that network continues to run, without acknowledging the existence of the hard fork since those blocks don’t conform to the rules of the network), and also gets the new coins. So when Bitcoin Cash split from Bitcoin, if a user originally had 10 bitcoins, she now had 10 bitcoins and 10 bitcoin cash coins. She could keep both sets of coins, or she could sell the set of coins that she didn’t want (assuming they are worth anything, with real buyers) and buy more of the ones she wants.
Users mostly chose to sell the bitcoin cash coins in that instance, and so bitcoin cash coins lost tremendous value compared to bitcoins. In addition, the Bitcoin Cash network had far fewer miners, and thus was less secure against 51% attacks. The divide has only grown since then; if just 1-2% of miners from the Bitcoin network decide to attack the Bitcoin Cash network and overwhelm its hash power today, they can do so.
What we know as Bitcoin or “BTC” is the blockchain that has not undergone any formal hard forks. It’s compatible with nodes that go back many, many years. Bitcoin Cash “BCH”, Bitcoin Satoshi Vision “BSV”, and other blockchains are the ones that are hard forks, meaning they split and were not recognized as bitcoins by the existing node network, but instead became their own thing.
Ethereum is different in this regard. What we know of as Ethereum or “ETH” today is a hard fork of a hard fork of a hard fork of a hard fork of a hard fork. It purposely updates via hard forks. In fact the minor altcoin ghostchain known as Ethereum Classic or “ETC” is the original Ethereum blockchain, at least out of the Ethereum blockchains that still exist.
In Ethereum’s early days, a massive flawed smart contract was exploited due to poorly-written code, and rather than let it play out as coded with investors losing money in their failed project, developers rolled back the entire blockchain with a hard fork, and due to broad community support, that hard fork became the dominant chain. The original blockchain where that change was not rolled back, mostly abandoned, became Ethereum Classic.
Since then, Ethereum has continued to hard fork a number of times to make updates, but those other chains that it forks from get abandoned without a name, since they are not as contested by anyone with significant resources like the Ethereum Classic chain was.
Since Ethereum updates via hard forks, and has “difficulty bombs” inserted into its code on the existing (pre-fork) chain, it gives developers a lot more control over the direction of the network than nodes. The Ethereum node network doesn’t realistically have the power to reject changes in the way that Bitcoin network nodes do, since a hard fork moves beyond their existing nodes anyway, and there are difficulty time bombs in Ethereum’s code. This gets users and miners to regularly agree to switch to new hard forks that developers come to consensus on.
Bitcoin proponents often criticize Ethereum’s level of centralization and ease of mutability. Ethereum proponents often defend it as necessary to change it into something better, to update faster. It’s a different set of philosophies, but it’s important to realize how different those philosophies are in the technical sense.
Stablecoin Custodians: Smart Contract Fork Deciders
Apart from difficulty bombs and things like that, there are powerful centralized forces in Ethereum that can dictate which hard fork is successful if a hard fork occurs. Seeing as how both intentional and unintentional hard forks happen with Ethereum quite often, that’s a relevant fact.
The Ethereum Foundation remains a powerful force for determining the direction of Ethereum. Consensys, which contributes to development and runs the Infura node infrastructure (which if it goes down basically brings down a large portion of Ethereum functionality as it did in November 2020 due to the chain split) and owns MetaMask (the key wallet application used by tens of millions of Ethereum users for DeFi apps and NFTs) is another powerful influence over the direction of the network.
But besides those two obvious centralization hubs, the often-overlooked sources of power are the largest stablecoin custodians. They have basically enough power at this point to dictate which Ethereum blockchain is valid, in the event of a hard fork. With $115 billion in assets between them, the two largest stablecoins have a lot of influence over Ethereum and other smart contract blockchains.
When a hard fork happens, stablecoin custodians cannot recognize both sets of tokens as redeemable for their money, since there are now twice as many total tokens (two full sets, one for each fork of the blockchain). They have to pick which blockchain is the valid one in their eyes, for which they accept redemptions of their tokens for money. And whichever one they don’t recognize as valid, has its DeFi and other stablecoin value eradicated. Most of the $100 billion in AUM locked up in DeFi protocols, the core lifeblood of Ethereum, is reliant on centralized stablecoins, as well as the stablecoins that are used by centralized offshore exchanges or that are being used for payments.
So, Ethereum users can’t necessarily fall back on their node network defense if developers and large entities want to change any of the rules of the underlying protocol (including money supply or any other variable). If a hard fork happens, and some large entities and stablecoin custodians acknowledge this new fork as the new main blockchain, then it doesn’t really matter what the existing nodes think. Their existing chain will almost certainly lose, with broken stablecoins and broken DeFi, and the new hard fork with new rules but functional stablecoins and functional DeFi, will win.
And it’s important to note that the stablecoins are known entities, and have dealt with legal action in the past. If governments where custodians are located (or where they can be extradited to, or where large countries have sway over) want to crack down on cryptocurrencies, they would have an easy time with smart contract blockchains. Governments could quickly seize custodian funds, blacklist all stablecoins, and render a large portion of DeFi insolvent across all smart contract platforms. Or they could enforce a hard fork with top companies and stablecoin custodians to create certain rules that the government wants the blockchain to have, like certain surveillance backdoors, or changes to other variables of the protocol.
A blockchain that is as self-contained as possible, like the Bitcoin network, is inherently more resistant to those types of attacks or centralization forces. There is no stablecoin provider, and there is no key wallet developer, that could direct the Bitcoin network in any significant way, especially when it comes to enforcing hard forks. There are some stablecoins that run on layers on top of the Bitcoin network, but they don’t run directly on the base layer of the protocol, and not in any size that is critical for the ecosystem.
That is why I classify bitcoins as being a form of money, while I classify most other cryptocurrencies as being a type of financial services equity, a more centralized platform with a pre-mine for the development of applications.
Smart contract blockchains are semi-centralized to varying degrees, demonstrably mutable, and therefore are political in nature. That doesn’t mean they can’t go up in price, and doesn’t mean they can’t offer functionality, but it makes them inherently different things than global immutable monetary assets, and so it’s useful to separate them into these two conceptual buckets.
How Important is Decentralization?
In bull markets, and at times with no regulatory crackdowns or drama, technical details don’t really matter.
Wall Street actually kind of loves DeFi in the tactical sense, because in aggregate they understand the idea of leveraging, liquidity management, exchanging, and arbitraging inefficiencies, and don’t really care about decentralization or technical details as much.
But for cypherpunks, sound money advocates, those who care about immutability and money supply assurances over a decade-long investment horizon, and those who care about securities laws, they notice.
It’s often said that a blockchain is basically just an inefficient database. Users are willing to trade inefficiency to ensure decentralization.
A blockchain, especially the truly decentralized variety, is a database that is small and tight enough that thousands or millions of entities around the world can store it on their local devices and constantly update it peer-to-peer using an established set of rules.
A fully-centralized database has fewer limitations, because it doesn’t need to be small and tight. A large service provider can have an utterly massive database, contained in a server farm. That can make things run very efficiently, but unlike with a blockchain, outside entities can’t directly audit it for content and changes, and have no control over it.
Your social media account is an item in a corporation’s database; it can be deleted or changed and you have no say in this. You have no way to audit what information they hold about you in their database. The same is true for your bank accounts, your criminal records, your health records, any cloud services you use, etc. Corporations and government entities have databases, and may at times choose to let you access those databases with limited permissions, or not. They are fully centralized, non-auditable, and easily changeable by the organization that runs it.
The killer application of a sufficiently decentralized database, is money. Money is a ledger, at the end of the day, and the more immutable it is, the better, at least for long-term storage. The ability to store value in a public ledger by simply saving or memorizing a number, and transfer that value to others internationally whenever you want, in a way that millions of other participants recognize and that no centralized entity can change or prevent or debase, is quite useful.
Smart contract layer one platform developers propose that there are many more potential applications that benefit from blockchain technology as well, besides just money. That remains an open question among cryptocurrency traders and investors; what are the other applications? Quick payments (e.g. stablecoins) seem to be an answer, and potentially things like settlement of securities, gaming, etc.
The biggest challenge with these proposals is that the more features you add to a blockchain on the base layer, the less “small and tight” it is, and therefore the less decentralized it tends to be.
The question then becomes, are there shades of partial-decentralization that people will accept, in exchange for more features that the database can offer? And can those partially-decentralized blockchains survive attacks, disagreements, and other tests over the long term?
Here’s another way to phrase the question. Since we know that there are use-cases for fully centralized databases (e.g. Twitter or Amazon Web Services), as well as use-cases for fully decentralized databases (e.g. the Bitcoin network), are there use-cases for a partially-centralized and partially-decentralized database?
If the answer is yes, then that’s basically the steel man argument for the existence of base-layer smart contract blockchains like Ethereum, Solana, Avalanche, Algorand, and more.
This set of hypothetical partially-decentralized databases wouldn’t conceptually compete with Bitcoin as truly decentralized immutable asset, but could they coexist alongside Bitcoin indefinitely as a semi-open operating system for apps that benefit from partial auditability or partially-decentralized control?
For example, if a database is controlled to some extent by a central organization, but it is open source and it is designed in such a way that its contents can be independently backed-up and audited in real time by certain high-performance external nodes, does that concept have an addressable market? Perhaps for payments and securities settlement?
And what about a federated database, meaning a database that requires the cooperation of several large organizations to change, or that requires proof of stake by large (and generally oligopolistic) entities, rather than a singular entity? Could that have long-term value?
I don’t have the answers to these questions, other than that with technology that currently exists or that is foreseeable on the horizon as of this writing, they’re clearly not suitable for truly decentralized global money in the same way that the Bitcoin network is. They might work for gaming, permissioned payment systems, trading, and that sort of thing, but time will tell if they can survive past the speculation phase and regulatory arbitrage phase that they are now in.
Overall, I view some of them as probably lasting for a long time if regulators allow them to, as information technology or financial services equities that pass the Howey Test and are therefore securities.
It’s also worth noting that smart contracts can exist as layers on top of Bitcoin as layer 2 solutions. In fact, they already do exist in that form, but those ones aren’t the dominant ones. The dominant ones are the versions that currently stand alone as layer one solutions, such as on Ethereum, Solana, and their various competitors.
Smart Contract Applications
So far, decentralized finance “DeFi” and non-fungible tokens “NFTs” are the two popular smart contract applications aside from just storing and transmitting value, that have gained significant market value on public blockchains. And both of them require additional complexity, and thus tend to cluster on blockchains such as Ethereum and Solana that, as discussed in this article, are more centralized than the Bitcoin network.
There is also a third category, decentralized autonomous organizations or “DAOs” that have gained a lot of press in recent months, even though they’re not on the financial scale of DeFi or NFTs yet. I’ll leave them for another article.
DeFi includes decentralized exchanges where users can trade various tokens between themselves, and includes decentralized platforms for leveraging tokens, meaning that users can earn yield by lending, or pay yield to borrow with collateral. Many of them still have centralized companies running them (e.g. Uniswap and Compound are both centralized VC-backed companies), but they do have open source code that sophisticated users can navigate without using the companies’ interfaces as long as the underlying blockchain is not compromised (and as previously discussed, those underlying blockchains do have centralized attack surfaces, so they are mutable to various degrees).
NFTs include things like digital art, unique game items, or digital movie tickets, that exist as unique items on a blockchain. Each category has some nuances about how they work. Digital art, for example, doesn’t actually exist on the blockchain, but rather there is a pointer on the blockchain that links to where the image is stored elsewhere. It is like owning a “signed receipt” from the artist of that image. Unique game items can include digital pets, or in-game items, or in-game land/property, and they can be sold to other players or even removed from the game and potentially accepted by another game that recognizes them.
The criticism of these applications so far is that they mainly revolve around speculation. Here is how I described DeFi back in my January 2021 Ethereum article, for example:
One of my concerns, when reviewing the biggest use cases for decentralized apps, is that a lot of the use-case is circular and speculative.
Ethereum is heavily used for decentralized exchanges of crypto tokens, crypto stablecoins that serve as liquid units of account for trading crypto tokens, and lending and earning interest on crypto tokens which is a practice that serves as a liquidity/borrowing source for traders of crypto tokens. To a lesser extent, it is also used for gamified ways to earn or trade various crypto tokens.
So, it’s a big operating system powered by crypto tokens, for the purpose of moving around… crypto tokens.
A healthy banking system in the real world would consist of people depositing money, and the banks making various loans for mortgages and for business financing, to generate real-world utility.
A speculation-based banking system, on the other hand, would consist of a bunch of banks taking deposit money, and then lending to speculators in the nearby stock market, along with technology providers that make this easier, and then what those speculators are trading mostly consists of shares of those banks, shares of those tech companies, and shares of the stock exchange, resulting in a big circular speculative party. The biggest use case so far for Ethereum is a decentralized version of that circular speculation-based system.
And data has shown that since I wrote that, it has become even more like this. According to the large blockchain analytics firm Chainalysis, DeFi is almost entirely a trading/leveraging/arbitrage environment for institutional-scale traders and professional whales, with individual retail traders strikingly absent:
Chart Source: Chainalysis
The same is generally true for NFTs. There has been a large frenzy of speculation around CryptoPunks going up, for example.
A key problem is that these types of NFT sets are pretty easy to manipulate because each one has a unique price, making it hard to establish what the real demand is. There are two easy scams that can be done with this asset that can’t be done with fungible liquid assets.
The first scam is to bid up asset prices and trick buyers into thinking those prices are real and to buy into it. It’s market manipulation, in other words. For example, a user can set up five different Ethereum addresses, and start trading around an NFT to themselves at increasingly higher prices. Outside observers don’t know that all these wallets belong to the same person and that this is literally just insider trading. This is only possible with a non-fungible asset; you can’t manipulate the price of an individual bitcoin or an individual ether on your own, you can only manipulate unique objects like for example CryptoPunk #9998. Then, with prices (seemingly) so high, some people want to get in on the momentum and buy the NFT, so the person who was trading among their own wallets finally sells the asset at a higher price to that unsuspecting newcomer. When that newcomer tries to sell the asset, he or she is unable to find other buyers who actually want to pay that price. They don’t realize that a lot of the liquidity and price-escalating transactions were actually just manipulation.
The second scam is to create a big loss to reduce tax liabilities in a fraudulent way. Again, you create several different wallets. One of them is linked to your real name and the others are anonymous. You buy an NFT with an anonymous account that you control for $200k, and sell it to another anonymous account you control for $250k. Then you sell it to your real-name account for $500k. Your real-name account then sells it to another one of your anonymous accounts for $200k, locking in a massive $300k “loss”. Your anonymous account can then potentially sell it for roughly what you paid for it, maybe $200k if the market hasn’t changed much since you began this trick. This is a useful tax “loss” (which wasn’t really a loss, since you secretly paid it to yourself) that can offset your real crypto capital gains from other trading areas.
To be clear, people who don’t enjoy spending time in handcuffs shouldn’t try those actions. This type of thing happens in traditional art as well but it can happen orders of magnitude faster in digital form.
And that’s not to say that all of the liquidity and price action is fraud. I don’t know how much is. It’s simply that, with the technology as it is, it is very difficult to distinguish what percentage is fraud and what percentage is real, and rising price action based on fraud can temporarily bring in real demand liquidity, making the difference between the two rather murky. This is not much of an issue for large cap liquid tokens but it’s potentially a big issue for non-fungible tokens.
There was an example back in October 2021 where CryptoPunk #9998 sold for $532 million. At first glance, this was the highest-value art sale of all-time. However, upon further analysis, it turns out that the buyer used a DeFi protocol to sell the asset to their own self, with a massive flash loan. They then tried to list it for $1 billion, but of course nobody wanted to buy it at that price. These are fake prices.
So far, the most popular NFT application for retail investors may be Axie Infinity, which is indeed played by millions of people in the Philippines and in many other countries globally, and for which the in-game currency is accepted by some outside merchants. However, the economics of that game are also inherently speculative because the majority of people can only make money if the number of new players continues to grow. A video game naturally runs into competition and a finite scale at some point, at which point the majority of participants would no longer be making money from the game.
Now, the argument from advocates in favor of these dedicated smart contract platforms is that it’s speculative here in the beginning, but that in time it will mature and be useful for more non-speculative utility related to a shared virtual economy. And I’m sympathetic to that view. After all, Bitcoin investors face similar allegations. In the early days, bitcoins were frequently used in the dark web, and today many people buy a little bit of bitcoin as a speculation to start with, and then as they learn more about it, they start viewing it more like a long-term asset to hold rather than speculate with.
One of the key smart contract applications that I think clearly is useful, is stablecoins.
From the user perspective, they’re generally a better way to handle fiat currency payments than, say, international wire transfers or large domestic payments. You can send payments, and clear them in minutes any time of the week. They will naturally face ongoing government regulation and be controlled and surveilled as part of the banking system in many cases, but it seems clear that they have utility for actual payments and will probably get increasingly incorporated into financial systems, either in the form of central bank digital currencies or private-but-highly-regulated stablecoin issuers.
This is simply due to automation and superior technology. When you send a wire transfer, the bank has to actively do something to process that transaction. And wires often get delayed or blocked or run into other problems as they flow between banks. From the users’ perspective, it’s often unclear which bank it got stuck in or who to call, and thus it often takes days to resolve. With stablecoins, it’s the opposite. The automatic nature of the blockchain allows for peer-to-peer transactions handled by software, including internationally and including with large amounts of money. The custodians are passive in that regard and let the technology work for them, and only act in the event that they want to blacklist some of their tokens for some reason that they detected.
In other words, regulated stablecoins allow for an automated peer-to-peer payment system, but with an overlay of surveillance and censorship based on know-your-customer and anti-money-laundering “KYC AML” laws.
Importantly, however, we see that stablecoins have been rather platform agnostic. Tether, for example, moved from primarily running on a layered solution on Bitcoin called Omni (red), to running on Ethereum (green), to increasingly running on Tron (blue).
Chart Source: Coin Metrics
Is Tron a better blockchain than Ethereum? No, it’s just cheaper. The less critical an application is, the cheaper people want it to be.
In other words, stablecoins as payment solutions tend to optimize for low transaction fees, and thus tend to concentrate towards efficient-but-centralized platforms. And all of the big stablecoins that underpin DeFi rely on centralized custodians anyway.
Will banks eventually just set up institutional stablecoin payment rails themselves, or devise similar solutions that are cheap and efficient? That’s essentially what Facebook has been trying to do with Novi and Diem; optimize stablecoins for actual payments rather than for trading crypto assets with.
It remains to be seen which platforms will be long-term stablecoin winners but it seems that they will trend towards rather centralized or federated networks to maintain low fees. The goal for many users isn’t really decentralization. Instead their goal is efficiency, with regulatory oversight.
Competing Base Layers, or Competing Second Layers?
If we put aside the current issues with DeFi and NFTs and grant for the sake of further analysis that smart contracts have a very large total addressable market, besides speculation and besides stablecoins, then a question becomes, who will the winning platforms be?
There’s an interesting narrative competition between Ethereum and Solana and Avalanche and others in the past several months. Ethereum is the established smart contract blockchain with a wide network effect, but with significant scaling problems and very high fees (and hence, small retail users are mostly absent other than speculating on the tokens by buying them on centralized exchanges), and is trying to transition from proof-of-work to proof-of-stake. Solana is a younger upstart VC-backed smart contract blockchain that comes with impressive scalability, but at the cost of more centralization. Avalanche proposes a complicated solution to trying to address this as well. Then there is Algorand and others.
DeFi and NFTs have thus begun to spill out from Ethereum onto these other smart contract blockchains. Many users are willing to sacrifice a bit of security for fees that are orders of magnitude lower.
Ethereum’s proponents often criticize (rightly) Solana as being too centralized, as their key defense for why Ethereum is better than Solana. But that puts Ethereum in a tight spot, because Ethereum’s proponents then have to criticize Solana as being too centralized, while also defending the fact that Ethereum has these centralized attack surfaces and greater complexity compared to Bitcoin. In other words, it has to justify what the right level of partial-centralization and partial-decentralization is, and that it has achieved this sweet spot.
As a result, smart contract platforms remain in the midst of a “Layer 1 war” with each other, as they battle for market share.
Meanwhile, the Bitcoin network has layers that can bring smart contracts to itself, and they keep getting more sophisticated. The Liquid sidechain, which is a federated sidechain that runs on the Bitcoin network, hosts NFTs including art, gaming tokens, stablecoins, and utility tokens. El Salvador announced plans to issue $1 billion of sovereign bonds on the Liquid network. Rootstock runs on the Bitcoin network as well, to bring DeFi and similar types of applications to the ecosystem. The Lightning network also hosts all sorts of proto-applications focusing on peer-to-peer data transmission.
These Bitcoin-based smart contract layers are currently much smaller than on Ethereum. This is partly from culture; bitcoiners tend to be holders more-so than speculators, tend to not want to trade other types of tokens as frequently, etc. But it’s also due to network effects and liquidity; Ethereum is still the dominant platform right now for pseudo-decentralized altcoin trading, leveraging, NFT speculation, and blockchain gaming, even though it is gradually spilling out onto cheaper smart contract platforms.
It’s unclear to me, looking 5+ years out, where this smart contract liquidity will wind up. Will it stay on Ethereum? Will it continue to gravitate towards even more centralized smart contract platforms like Solana and Avalanche and so forth, so that we have an increasingly diluted multi-chain smart contract world? Or will speculation subside and the most utilitarian use-cases find their way back to layers on top of Bitcoin due to an appreciation of Bitcoin’s more solid base layer?
Ultimately, it partially depends on what governments want. Smart contract platforms with centralized attack surfaces can only exist at the pleasure of the government, so it comes down to how much regulatory crackdown they get vs how much regulatory approval they get.
In a relatively non-hostile environment, smart contract platforms tend towards commoditization, competing based on price rather than quality. Liquidity trends towards whatever is cheap, centralized, and with sufficient critical mass. There are network effects for liquidity, but these are somewhat offset by high fees, which kind of serve as anti-network effects.
In a more hostile environment with regulatory crackdown or other attacks, then the chains that are too centralized are likely to find it impossible to operate, whereas chains that make throughput sacrifices to maintain some degree of decentralization are able to operate to some extent. Liquidity would naturally need to flow towards the one or a small number of chains that are able to operate in that environment.
My overall base case is to see a number of smart contract platforms continue to operate in increasingly-regulated ways, constantly fighting for market share.
Peer-to-Peer, Without DeFi
When the Bitcoin network was originally created, there were no exchanges. If people wanted to buy or sell bitcoins, they had to make individual arrangements. There would naturally be some organized meetups to make this easier, and the industry eventually formed centralized exchanges.
But at it’s core, it is peer-to-peer technology. If you and I meet in person, I can agree to send you a fraction of a bitcoin from my bitcoin address to yours, in exchange for cash or any other good that you hand to me, and we can do it at a coffee shop.
For people who prefer to avoid centralized exchanges, there are various peer-to-peer technologies that make this easier than an in-person meetup like that. Bisq, Hodl Hodl, LocalBitcoins, and Paxful are all various ways to do peer-to-peer bitcoin exchanges, and each have different trade-offs but don’t require external tokens.
An escrow multi-sig platform, for example, can serve as an independent third party. Buyers and sellers can enter into a 2-of-3 multi-signature contract online, where the seller puts in their bitcoins, and they only get released when the payment from the buyer is made. A third party holds the third key of that contract, which ensures the bitcoins are only released if both parties are happy, and can be an arbiter of disputes to accept proof if one of the parties is unhappy, before finalizing the transaction.
Nigeria cut off crypto trading from its banking system a while ago. They didn’t make owning or trading cryptocurrencies illegal (that’s very hard to enforce), but instead they went with the simpler move of severing crypto from any formal connection with their domestic banking system. You can’t take Nigerian fiat currency and easily send it to a crypto exchange to buy bitcoins, in other words.
In order to understand the game dynamics of that decision, realize that Nigeria has persistent double-digit inflation and does not want capital flight out of its banking system into a sound money digital currency that its citizens can easily transact with, but also doesn’t want to spark unnecessary social unrest by banning it (since it’s very popular) and wants its citizens to be able to receive bitcoin payments from abroad. Because Nigeria has a lot of good programmers and graphic designers that foreigners are happy to hire and pay bitcoins for with a population of well over 200 million, Nigeria has little incentive to put resources into going door-to-door and make sure Nigerians aren’t using bitcoins.
But the point is, individual Nigerians needed to find alternative ways to transact with bitcoins. And despite that, Nigeria has some of the highest adoption of bitcoin usage, ranked at #6 worldwide on a per-capita basis. They often use peer-to-peer trading using Paxful and LocalBitcoins to send and receive bitcoins peer-to-peer. And they use Telegram groups and other types of coordination to exchange fiat currency for bitcoins. They don’t use DeFi blockchain platforms en-masse. DeFi, with its high fees, is mainly for large institutional speculators, arbitrage players, whales, etc.
DeFi, thus far, is primarily for speculation. When people living in countries with GDP per capita of $2k USD, $3k USD, or $4k USD are interested in bitcoin, they don’t pay hundred-dollar fees on Ethereum to mess around with NFTs or crypto trading or leveraging. They form groups to arrange peer-to-peer bitcoin buys/sales, or they explore the cheapest (and often more centralized) smart contract platforms.
The notable exception to this general observation is gaming. As previously-mentioned, Axie Infinity is very popular in the Philippines, but a lot of that involves people grinding to get income from the game and the economics of that only work as long as the game continues to grow. If new player funds don’t continually pay out existing players incomes, then the game is subject to a collapsing user base unless it’s inherently fun enough for most users to invest in heavily, despite no longer getting net income out of it.
Protocol or Operating System?
Ever since Satoshi Nakamoto created Bitcoin, countless attempts have been made to improve on his design. For the major categories:
-People debated increasing the block size in exchange for nodes being harder to run, leading to greater centralization, and made new coins based on that.
-People debated reducing the block times, in exchange for less network stability, and made new coins based on that.
-People chose to sacrifice some degree of auditability for greater privacy, and made new coins based on that.
These coins consistently failed to hold even 5% of Bitcoin’s market capitalization. The wisdom of the market has decided over rather long periods of time now that it’s not interested in them, at least outside of niche circumstances.
Meanwhile, the Bitcoin network itself continues to update slowly on the base layer via soft forks, meaning it only makes backward-compatible upgrades, and only when there is overwhelming consensus to do so. And it continues to update quickly on second layers, on side chains, and with hardware and software providers in the surrounding ecosystem, that don’t affect the base layer. Some of those upgrades can make using the Bitcoin network faster, with more throughput, with more features, and/or with more privacy.
The big bullet point that the market is still deciding on, though, is this one:
-People debated adding more capabilities to blockchains at the base layer in exchange for greater centralization and attack surfaces, and made new coins based on that.
So, a big topic that the market is still assessing, is whether these partially-centralized smart contract platforms have a big role to play alongside the Bitcoin network, or if they’ll eventually stagnate as previous altcoin speculations have.
I’ve seen a number of arguments for what the cryptocurrency field will look like after a long enough timeline. Ultimately, it comes down to whether the space evolves more like protocols do, or more like operating systems do.
Protocols tend to be winner-take-all outcomes, and then hold their position for a very long time with like 90% market share or more. TCP/IP is the protocol that the internet runs on and was developed in the 1970s. SMTP is the protocol for email and was developed in the early 1980s. Ethernet is the protocol for networking and was developed in the early 1980s. USB is the protocol for interfacing, and was developed in the 1990s.
In ten or twenty years, we’ll still be running on most or all of these, and they upgrade over time. These protocols all had competition initially, but most people today can’t name those competitors.
Operating systems tend to be oligopoly outcomes, rather than winner-take-all outcomes. Multiple operating systems can coexist, each with their own network effect and areas of preference, but only a handful of them can realistically have widespread appeal with critical mass of developer adoption. The same tends to be true for social network platforms, as well as financial exchanges.
Some people propose that after a sufficient maturity of the field, one blockchain will dominate the field, with the argument that they are protocols, and one protocol (e.g. Bitcoin) will win.
Other people propose that the outcome will instead look like operating systems, with a small number of persistent large players. Even if one player might have 30, 40, 50% or more of the market, it won’t have 90%+ according to this view. A subset of this argument further proposes that Bitcoin and smart contract platforms like Ethereum aren’t even really competing for the same market, and thus can be grouped separately with only moderate overlap.
I don’t have complete conviction on how that will end up. It’s clear that Bitcoin won as far as decentralized proof-of-work blockchain money is concerned. And I think people underestimate the total addressable market size of that concept.
Aside from that, will there be persistent large smart contract platform, or will they one day fold into Bitcoin as layers on top if it? And to the extent that they remain as independent layer-1 smart contract platforms, to what extent will they dilute each other and fracture into commoditized highly-centralized low-cost blockchains? The market is still sorting these questions out.
Ultimately, my base case between the protocol outcome and the operating system outcome depends on the level of regulatory crackdown.
For the protocol outcome, I can envision that smart contract platforms either get hit hard in their attack surfaces (draconian regulatory crackdown, for example), or they crumble under the weight of their circular speculation aspects. Meanwhile, Bitcoin has a decentralized base layer and the ability to build smart contract applications on top of it on other layers, and it can pull that value in over time as other blockchains run into problems.
For the operating system outcome, I can envision that Bitcoin retains the dominant market share of global money and collateral in the digital asset space, with additional layers of complexity built on top of it as well, but that separate large smart contract platforms exist as regulated platforms for cheap stablecoin processing, crypto gaming, altcoin trading, NFT speculation, securities settlement, and other applications. These would basically be equity securities.
Final Thoughts: Always Consider Tradeoffs
There are about 15,000 cryptocurrencies in existence as of this writing, as identified by CoinMarketCap.
Bitcoin’s share of the the total cryptocurrency market changes over time, but for example it currently has about the same share of the market (around 40%) now against these 15,000 coins as it did four years ago against only 1,500 coins. So, altcoins have mostly diluted each other.
The way altcoins market themselves, generally, is to highlight the shortcomings of Bitcoin as though it were old tech or “Boomer coin”, and then explain how they are better than Bitcoin.
When you dig into them, however, it turns out they are making tremendous trade-offs in one area to achieve additional capability elsewhere. They are sacrificing some degree of security, decentralization, auditability, and so forth, in order to achieve things like more features, more speed, or more throughput. And now the same thing is happening to Ethereum; newer smart contract chains offer greater efficiency in exchange for more centralization, and criticize Ethereum for not sacrificing more decentralization to scale faster.
Satoshi Nakamoto picked his variables very carefully. Each one has been debated and tested.
Governments are good at cutting off the heads of a centrally controlled networks like Napster, but pure P2P networks like Gnutella and Tor seem to be holding their own.
-Satoshi Nakamoto, November 7, 2008
When truly better ideas come along for a small part of the protocol after years of proof, Bitcoin developers, supported by the users, eventually tend to incorporate them into Bitcoin with a consensus soft fork such as the Segwit and Taproot updates.
People often think of cryptocurrencies as one big similar asset class but for the most part, proponents of other blockchains are often the most vocal critics of the Bitcoin network, as they attempt to market their coin over bitcoins. Meanwhile, bitcoin enthusiasts are among the crypto ecosystem’s largest critics, and tend to highlight the scams, hacks, wash sales, and centralization problems that are common among the altcoin cryptocurrency space.
Crypto exchanges with numerous coins have an incentive to get you excited about new coins, because they make money from trading volumes. Even if it’s just meme-coins like Doge or Shiba Inu with briefly-lived spikes, they want to get you in on the action, especially near the top of the spike when enthusiasm is high. Their financial incentive is for their users to hold a large number of coins, and trade those coins frequently, and are happy to highlight whatever coins happen to be popular at the moment. In that environment, it’s the house (exchange) that wins either way.
To the extent that an investor chooses to speculate in digital assets other than bitcoins, they should always be able to answer the question “what are the trade-offs?” for one protocol compared to another before they decide to buy. Overall, I conceptualize bitcoins as monetary assets, and smart contract platform tokens as equity securities.
Each person has their own reasoning and penchant for speculation or long-term investing, but make sure you understand what you’re getting into when you venture into blockchains other than the Bitcoin network, rather than buying into the marketing without verifying each claim.