Carndano Is Not As Decentralised As You May Think

There has been a lot of discussion in the community of late regarding the decentralisation of Cardano. I thought that as a whole, it would be interesting to analyse a little deeper the specifics of that decentralisation.

Of course, decentralisation is critical in any digital currency. However, in traditional PoW systems, much of the subsystems are opaque. The decentralisation of Hash power is potentially easier to grasp than the decentralisation of pools and delegation

Second quotations In between the short code . For example, the bitcoin security budget is paid by the network to miners who provide a service. However, in PoS systems, the security budget is both a service fee and interest paid.

Let’s look at how the network currently works to see if we can break it down into its separate parts. In the current Cardano model, delegators stake their ADA with to pools. Pools are then randomly selected proportional to their stake to produce blocks. All pools at saturation (₳64Mn) should have an equally random chance of becoming block leader. However, participants assign delegation to pools in a free market and are heavily defined by external benefits rather than the capacity to secure the network.

The underlying issue with this system is entrusting the network’s security to those who have the most. Those with the most capital control the pools. Pools with the most marketing and external value add control the network. This makes Cardano ultimately beholden to its egalitarianism. The more egalitarian the distribution of the delegated asset, the more secure the network is. The ultimate example of this is if, in a PoS network, each user only possesses one token. Obviously, this could never practically work, and even if network designers could manufacture this situation, it certainly wouldn’t last long.

This seems to be on the minds of IOHK, who published the article Cryptocurrency Egalitarianism: A Quantitative Approach in 2019.

Proof-of-stake is often criticised for its lack of egalitarianism. The rationale is that, in proof-of-stake, the more money one stakes, the more money one generates. Thus, the rich get richer, which is precisely the opposite of egalitarianism. Additionally, in proof-of-stake systems, the money owners could constitute a closed, rich club, refusing to share the assets with any outsiders. ¹

The article concludes blockchains, both proof of work and proof of stake, are unfortunately plutocratic. ^[“Thus, as illustrated in this paper, blockchain systems are, for the time being, plutocratic. Whether decentralised decision making in which each human is allocated one vote is possible remains an open question.] Contrary to popular belief, Lars et al. stipulate that PoS systems are not inherently more so. Similarly, PoW networks are not more egalitarian simply by their design. This could present an issue for Cardano, whose chain security is determined by its effective egalitarianism.

This issue could perhaps be solved on the pool side. By enforcing strict single pools and generous pool limits, pools could take on the distributed security of a plutocratic PoS network. IOHK published a separate research article in 2020, which investigates the security tradeoffs of various parameter changes. This article shows how both a nash equilibrium and a sibyl resistant network are possible with the correct parameters. The proposed reward sharing scheme with the k and α parameters. Parameters users of the network will know well.

Our main result is the introduction and analysis of a novel reward sharing scheme that is parameterised by (1) the desired number of pools k, and (2) a Sybil resilience parameter α. The two parameters can be selected to fine-tune two desirable properties of the resulting configuration. The primary property is decentralisation and fairness ²

This sounds amazing. Defining mathematical principles to set parameters at a point determined by game theory to optimise decentralisation and fairness. However, there are some issues. This article and the network have some critical distinctions. The most glaring is that the article assumes each player can create only one pool ^[we have assumed that each player can create at most one pool and that the utility that an inactive pool gives to its members is zero. https://arxiv.org/pdf/1807.11218.pdf]. As we know, it is not only possible for a single entity to create multiple pools but is, in fact, expected. This fundamental change to the protocol ultimately makes the research conclusion inconclusive. This specific point is noted in the article Cryptocurrency Egalitarianism: A Quantitative Approach from Iohk ³

We should note that variations of proof-of-stake, such as “delegated proof-of-stake,” may not be perfectly egalitarian, since the delegates, i.e., the leaders of the stake pools which are formed, typically earn extra profits for managing the stake pools

This is defiantly an issue then. Cardano’s security needs to be egalitarian to be secure but, as shown above, is not.

Focus

When discussing decentralisation, it is important to frame it against a goal. For some, decentralisation is about decision making; for others, it is purely an expression of security. When discussing the threat of a Sybil attack, security is most at risk, so that is the framework we will use as a guide.

In the article Quantifying decentralisation Balaji S. Srinivasan and Leland Lee breaks the Bitcoin network down into its subsystems. All subsystems can present a risk to the network at large. We must analyse them all to give a practical overview of the health of decentralisation. Of course, some subsystems are more vital than others. I will be taking a similar approach.

a decentralised system (like Bitcoin) is composed of a set of decentralised subsystems (like mining, exchanges, nodes, developers, clients, and so on). ⁴

With Cardano, many subsystems, the leaders, wallet clients, nodes and developers could all be points of concern. However, as demonstrated above, the two core subsystems are delegators and pools. They are the two also most hotly discussed and will focus on the rest of this article.

Delegators

As shown above, all proof of stake systems are inherently plutocratic, and Cardano is no different. Just how unequal it is was did surprise me, however.

Just 468 individual accounts control 51% of ADA ^[Calculation made from data scraped from adastats.com], and therefore the security of the network. To put that into perspective, ~3 Million total accounts are currently active on the network. Potentially worst is the fact that the top 1% of Cardano holders own 91.56% of ADA ⁵.

We can quantify these numbers further using methods derived from the research paper “Measuring Decentralisation in Bitcoin and Ethereum using Multiple Metrics and Granularities” by Qinwei Lin et al. ⁶. The report identifies three key metrics for decentralisation:

1.  The Gini Coefficient investigates the distribution of various entities, including the underlying asset.

2.  The Nakamoto Coefficient, widely known, looks at the potential for Sybil and other attack vectors.

3.  Shannon Entropy quantifies the inherent randomness of the protocol design.

The latter is not so relevant for delegators but could further investigate the protocol.

The Gini Coefficient is 0.96. With 91% of the accounts holding just one per cent of the total ADA, it is not surprising that the Gini Coefficient is also high. As for the Nakamoto Coefficient, we mentioned it earlier. It is 468. ^[This also does not take into account multiple accounts owned by a single entity]

These numbers on their own. So let’s compare them with some other popular blockchains. ⁷

| Network | Gini |

| — | —- |

|Cardano | 0.96 |

|Bitcoin | 0.65 |

|Etherium | 0.98 |

|Doge | 0.82 |

|USA | 0.85 |

|Ukraine | 0.95 |

|Australia | 0.65 |

|Total Global Wealth | 0.80

As we can see, the Cardano network has one of the worst (highest) Gini Coefficients of all the blockchains noted. While not as high as Ethereum’s, approaching 1, Cardano’s Gini is remarkably higher than fellow UTXO chain, Bitcoin. Bitcoins Gini is similar to Australia’s and lower than the Global Wealth Gini.

It is essential to ensure that the distinction between PoW and PoS is apparent here. As mentioned above, Cardano is secured by its egalitarianism. This limitation is not the case with more traditional PoW chains. A Gini of 0 in a PoS network would indicate perfect decentralisation where all participants own an equal share. A Gini of 1 is perfect centralisation, where there is a single participant. Cardano has a Gini of 0.96.

Pools

There are currently 3167 pools registered and operating on the Cardano network. This is the number ofter cited by the community. However, the most critical metric for pools, and the only one which is relevant to a discussion on decentralisation, is how many blocks they have produced. Only 1029 pools, just 32%, are currently producing blocks. Of these 1029, many are in multi-pool groups.

The Nakamoto Coefficient of the Cardano network is currently around 28 ^[Blockchain insights] up from a previous local low of 22. This indicates that 28 entities control 51% of the network. In other words, based on what we found above, 468 accounts delegate to just 22 entities. This is what secures the network as a whole. It is important to note that this is not 468 people staking with a specific set of pools but rather just an approximation of risk.

We can again compare this against other chains.

|Network | Nakamoto Validators| Nakamoto Currency |

|—|—|—|

|bitcoin | 4 | 4652 |

|Etherium | 3 | 100 |

|Cardano | 28 | 468 |

In this table, we can see that Cardano is far healthier than its competitors. One interesting note is that the Nakamoto for mining in PoW networks is calculated using pools as single entities.

Conclusion

The concerns the community has regarding decentralisation are warranted. Cardano’s decentralisation is a multi-faceted issue with specific metrics flying under the radar. If Cardano is to become the global financial settlement layer, as some think is possible. The network must begin to make structural changes and implement further research. I have shown in the above that Cardano, to be secure, must be egalitarian, and yet is not.

The community in its commendable and insatiable drive for decentralisation may be distracted by MAV and pool centralisation and miss the much larger and systemic issue that the network faces.

Future work

Another element that I feel requires more analysis is the combination of pool control. Pledge is an element of pool structure which is often touted as a positive element of pools. However as shown above pool entities controlling both pools and staking ADA may be dispaportioanty increasing the risk to the network.

Thanks

I would love to extend a massive thanks to

• @m₳tiwinnetou creator of Cardano Blockchain Insights [Pool: CRFA]

• @Cardinians creators of adapools.org [Pool: CRDN]

• The community team behind adastat.org [Pool: STAT]

These guys do a fantastic job maintaining some of the best data and metrics on the health of Cardano. Without these data sources, Cardano could not be as transparent and open as it is. I used all these services extensively. I wanted to specifically thank @m₳tiwinnetou for hitting me up in Twitter DM’s and giving me some advice on Cardano Data.

References

Brünjes, Lars, et al. Reward Sharing Schemes for Stake Pools. arxiv.org/pdf/1807.11218.pdf. Accessed 20 Feb. 2022.

Karakostas, Dimitris, et al. Cryptocurrency Egalitarianism: A Quantitative Approach. arxiv.org/pdf/1907.02434.pdf. Accessed 20 Feb. 2022.

Karir, CFA, Vishal. “Measuring Decentralization in Crypto.” Ethos, 15 May 2018, www.ethos.io/blockchain-finance-measuring-decentralization.

Lin, Qinwei, et al. Measuring Decentralization in Bitcoin and Ethereum Using Multiple Metrics and Granularities. arxiv.org/pdf/2101.10699.pdf.

Nguyen, Cong T., et al. “Proof-of-Stake Consensus Mechanisms for Future Blockchain Networks: Fundamentals, Applications and Opportunities.” IEEE Access, vol. 7, 2019, pp. 85727–45, https://doi.org/10.1109/access.2019.2925010.

Rosenthal, David. “Gini Coefficients of Cryptocurrencies.” Dshr.org, 25 Dec. 2019, blog.dshr.org/2018/10/gini-coefficients-of-cryptocurrencies.html.

Sai, Ashish Rajendra, et al. “Characterizing Wealth Inequality in Cryptocurrencies.” Frontiers in Blockchain, vol. 4, Dec. 2021, https://doi.org/10.3389/fbloc.2021.730122.

Srinivasan, Balaji S. “Quantifying Decentralization.” Medium, 31 Oct. 2017, news.earn.com/quantifying-decentralization-e39db233c28e.