One of the arguments against scaling Bitcoin and Ethereum on-chain is the so-called blockchain trilemma. It’s a theory popularised by Vitalik Buterin which states that blockchains can only do two out of three at a time: decentralisation, security, and scalability. So, supposedly, in order to preserve decentralisation and security we have to sacrifice scalability.
It sounds smart as it reflects the more common, known to everyone, “fast, cheap, and good” triangle. Indeed, you wouldn’t expect something of a high quality made fast for a penny. So, naturally, exploring the trilemma, you’d say that you can’t have a blockchain which scales and remains decentralised. The only problem is that this trilemma is a total hoax. Let me explain why.
The first issue is measuring the sides.
You can measure “fast” in seconds, “cheap” in bitcoins, and “good” in the number of stars reviewers leave (or in tips, whatever).
But how do you measure “decentralisation”? There’s a popular approach called the Nakamoto coefficient which measures how many different entities (miners, validators, etc.) are needed to disrupt a blockchain. But it can be quite bad disregarding any scaling. As of the beginning of 2024, just two mining pools (Foundry USA and AntPool) control 51%+ of the Bitcoin’s hashrate and can (but not really economically motivated) pull off a 51% attack. Does this mean that Bitcoin is already not decentralized? Bitcoiners would say “no”.
For most people, measuring “decentralisation” is more of an intuitive feeling. Some people simply argue that Vitalik controls Ethereum single-handedly. The important point is that there’s no “consensus” on how to measure “decentralisation”.
It’s even harder with “security”. Generally, I tend to say that “security” is measured in money. Or relative money. It makes sense to put your diamonds into a safe you have to purchase first, but it doesn’t make sense to buy a tank to protect your $10 bill (unless you’re Scrooge McDuck protecting your lucky sat).
But in decentralised systems, “security” highly coincides with “decentralisation”. Does it really matter that it costs $1 trillion to 51%-attack Bitcoin if there’s a single miner capable of doing this? Decentralised blockchains are not secure if they’re not decentralised enough. Now we have a little feedback loop here as we can’t measure “security” without first properly measuring “decentralisation”.
“Scalability”, on the other hand, can be easily measured in something like “transactions per second”.
Now, the theory behind the trilemma is that if we further scale a blockchain, we’ll lose in decentralisation or in security. But is it really a solid theory if nothing can be really calculated?
The second issue is that it doesn’t work backwards.
One would expect the trilemma works as a two-way street. If we scale a blockchain, we lose in decentralisation. If we descale a blockchain, we gain more decentralisation. But is it so?
Well, indeed, if we increase Bitcoin’s block size to 1 exabyte, there will be some issues with running nodes, which well affect our “intuitive” decentralisation measurement. It will become even more hard to run a node, so maybe we’ll be down to just one mining pool controlling 51%+ of the hashrate. As if having two pools wasn’t bad enough.
But what about decreasing the block size to, say, 1 kilobyte? We’d descale Bitcoin that way, it’d become easier to run a node, but what about security and, in turn, decentralisation?
Miners won’t be really able to include anything besides a coinbase transaction. They won’t be able to earn in fees opening Bitcoin to an even worse scenario of the security budget issue. Maybe they’ll be able to fit in a single clearing transaction for some L2, but no user will be able to use Bitcoin non-custodially anymore on L1. This is neither “secure” nor “decentralised”.
The third issue is that it doesn’t work at all.
Scaling to 1 exabyte or descaling to 1 kilobyte are, of course, extreme examples. But what about something more cautious?
How would increasing the block size by, say, 400% affect decentralisation? Would it drop decentralisation (however you measure it) by 400%? Would it even drop by 1%? How many of the current Bitcoin miners earning tens of millions of dollars a month won’t be able to upgrade their node servers? Not to mention that increasing the block size doesn’t even require a lot of hardware upgrades. And what about the tech progress?
SegWit made it possible to mine 4 MB blocks, a 100% increase from 1 MB. Does it mean that on August 24th, 2017 Bitcoin became 400% less decentralised, or 400% less secure, or maybe 200% less decentralised and 200% less secure at the same time? No, we didn’t really see any drop in the node count (mining and non-mining) or the miner revenue.
You can’t plot a mathematical function here. The trilemma is nothing but a bunch of buzzwords. It only works at extremes, but do we really need 1 exabyte blocks?
Introducing The Reverse Blockchain Trilemma
In all fairness, it’s important to note that blockchains don’t exist in a vacuum. They exist on a competitive market for P2P electronic cash. Blockchains may have their own advantages (whether it’s a first-mover advantage, most PoW, most PoS, cool features, etc.), but they compete with each other.
Now, imagine someone wants to run a new business (or even a new protocol) on Bitcoin. Unfortunately, as Bitcoin is unable to scale its L1, this someone decides to move to another blockchain, say, Ethereum. What happens?
- Bitcoin’s decentralisation is not increased, as this person won’t be running a node.
- Bitcoin’s security is not becoming better, as they won’t be paying fees to Bitcoin miners.
- Ethereum is the winner here: decentralization is increased, as they will be running an Ethereum node, and validators will earn more money by collecting their fees.
Not scaling in time is a blown opportunity to increase both decentralisation and security. This has already happened in Bitcoin vs. Ethereum: by a lot of measurements Ethereum is doing better than Bitcoin. Ethereum users pay more in fees, there are more Ethereum nodes than Bitcoin nodes, and so on. But the very same problem is looming over Ethereum as it doesn’t scale anymore either.
The Reverse Blockchain Trilemma can be defined as this: “You won’t be able to have all three: not scaling L1 to fit all users, the most secure blockchain, and the most decentralised blockchain”.
Twitter discussion: https://twitter.com/nikzh/status/1745787103344988619
Thanks @ddadybayo for a review.