Many blockchain networks claim to be decentralized, but few have concrete metrics to substantiate this claim. Often, these networks cite their node count as a primary measure of decentralization. However, is node count alone sufficient?
If node count is indeed a key metric, it's essential to differentiate between types of nodes. For instance, in a network with separate validators and other nodes, a high node count might not equate to true decentralization if only a small subset are validators. Comparatively, a network with an equal number of nodes where all participate in validation would be more decentralized.
Additionally, the hosting locations of nodes matter. If a network's nodes are concentrated in just a few data centers, it represents high centralization. Whether there are 10 or 10,000 nodes, the system is vulnerable if two data centers failing can take down more than 50% of the nodes.
Decentralization is challenging to measure, but one metric attempts to quantify it: the Nakamoto coefficient.
The Nakamoto coefficient, proposed in 2017 by Balaji S. Srinivasan, former CTO of Coinbase and General Partner at Andreessen Horowitz, has been used to measure decentralization in networks like Solana, often giving it a higher value compared to Bitcoin or Ethereum.
What is the Nakamoto Coefficient?
If decentralization is viewed as binary, no measure is needed—it's simply a yes or no. However, the Nakamoto coefficient operates on the premise that decentralization is a spectrum. It aims to measure the extent of decentralization, assess how changes affect it, and optimize algorithms to enhance it.
The coefficient's concept is straightforward. Every decentralized system comprises subsystems. By identifying how many entities need to be controlled in each subsystem, one can gauge the network's effective decentralization. The more entities required for control, the higher the decentralization.
The coefficient's calculation is based on economic principles like the Lorenz Curve and the Gini coefficient.
The Lorenz Curve
Developed by Max O. Lorenz in 1905, the Lorenz Curve illustrates wealth distribution in populations, highlighting inequality similar to centralization. It plots cumulative income percentages against population segments. Perfect equality would be a straight 45-degree line. More unequal distributions push the curve away from this line.
Economists use the Lorenz Curve to calculate the Gini coefficient, measuring inequality. The area between the Lorenz curve and the equality line, expressed as a proportion of the area between the absolute equality and inequality curves, gives the Gini coefficient (A/(A+B) in the example). A higher Gini coefficient indicates greater inequality, akin to centralization in networks.
Measuring Decentralization
The Nakamoto coefficient combines the Gini coefficient with subsystem control to determine network decentralization. Balaji identifies six subsystems for blockchain networks:
- Mining (by reward)
- Client (by codebase)
- Developers (by commit)
- Exchanges (by volume)
- Nodes (by count)
- Ownership (by addresses)
The critical threshold for compromising subsystems is typically 51%, though it can vary. For PoW chains, controlling 51% of computing power can compromise the network. For exchanges, a higher threshold might be needed to affect liquidity.
Nakamoto Coefficient for Mining & Staking
Consider Bitcoin's hash rate distribution:
To compromise the network, 51% of the hash rate must be controlled. Currently, four mining pools (Antpool, F2Pool, ViaBTC, Binance) need to combine efforts to achieve this, giving Bitcoin a Nakamoto coefficient of 4. Ethereum's coefficient is even lower, with three pools controlling 61% of the hash rate.
For PoS networks like Solana, controlling 33% of the stake suffices. Solana's website states it takes 19 validators to compromise the network, resulting in a higher coefficient than Bitcoin or Ethereum. However, PoW networks' hash rates are more fluid, whereas unstaking in PoS networks takes time.
Nakamoto Coefficient for Ownership
Examining ownership by wallet addresses reveals more about decentralization. Ideally, no single entity should significantly influence the market with their holdings. However, large holders (whales) are common in crypto. For example, SHIB has 982,000 addresses, but 14 hold 68% of the supply, with one address alone holding over 40%. This centralization results in a Nakamoto coefficient of four for SHIB ownership.
A caveat: anyone can create multiple wallets, potentially making distribution seem more decentralized than it is.
Conclusion
The Nakamoto coefficient is a valuable measure for comparing blockchain network decentralization. It encompasses various dimensions, including node count, computing power distribution, ownership, and trading venues.
In all examples, the coefficient is relatively low. A truly decentralized protocol would have a coefficient in at least the five digits.
Well written, nice1
