Decentralization should be an easy concept to grasp, there is no central authority and there are governance mechanisms in place to ensure the smooth running of the decentralized organization or system.
If we look at blockchains we see that in practice it is harder to achieve Let’s take Bitcoin where anyone can set up a node and mine, in theory, the Bitcoin network should be nicely decentralized, yet we see a concentration of miners in China which represent more than 75% of the mining power. The concentration of mining in China is due to the combination of being close to where the hardware is produced and the cost of electricity. The concentration of miners in one country may not be an issue for the running of a blockchain, but it begins to look like centralization.
The Proof of Stake (PoS) mechanism is an alternative to the Proof of Work (PoW) mechanism used by Bitcoin and replaces the scarcity of computing power with the scarcity of tokens by requiring the node operators to stake tokens (a form of escrow) which reduces the supply making it expensive to buy up tokens to attack the chain.
We see networks running PoS with several hundred validators and from a distance, it looks fully decentralized; validators are selected based on their voting power (stake) and actively seek delegators to stake their tokens. Dive a little deeper and things may not be as they seem. There are few issues with the model, delegators are supposed to do due diligence on the validators they delegate their stake too, and ensure they run a good operation, in practice, there seems to be a trend for delegators selecting on size of validator and amount staked. There has also been the “zero commission attack” where a validator offers no commission on the returns they generate to attract delegators and they can do this as they have a reasonable stake in their own right and need the extra stake to achieve higher voting power (and thus income). We also see a validator with purchasing power where they buy up a lot of tokens (normally in the bootstrapping phase) and use it to build authority. This leads to a concentration of power with a handful of validators having disproportionate voting power, we can see from the leader board in Cosmos the top 8 validators have 35% of the voting power.
The EOS blockchain has been mired in discussions about the block producers colluding, and it is symptomatic of a design that has a small number of block producers. The collusion of validators can turn into a cartel where the validators run a chain for their benefit which has been seen in Lisk
There is also an issue around token distribution, especially in the context of PoS, it becomes an issue if “insiders” retain a significant proportion and smells of centralization. We see, for example, in Polkadot where over a third is held by insiders and Parity Technologies This combined with the model of one set of validators running a central hub with the parachains renting security causes issues around centralization.
While a visionary leader in a blockchain is an asset in driving the direction of the blockchain, it can also be an issue where a single person has disproportionate influence and thus introduce centralization (especially if the leader is a significant token holder).
What do we look for in a decentralized network?
The key elements are that the network must have been designed to be decentralized, easy to say, harder to establish.
This is not a definitive list but guidance on what to look out for when assessing whether a blockchain is decentralized or not.
- Reading the various whitepapers of blockchain projects, especially where consensus is covered, there are many flavors of consensus mechanisms, and these will give the first indication of the intent of the blockchain and how they are designed.
- Find the tools that show voting power for the relevant blockchain. A quick calculation will show the levels of concentration in the validators.
- Check how many nodes are active, the more the better, especially when voting power is equally distributed.
- Check the number of tokens held in addresses as an indication of the concentration of tokens that may be used in voting or in staking. A small number of addresses with a significant quantity of tokens could be significant in assessing centralization, although it must be noted that one person could control many wallets and that is much harder to spot.
- Delegated Proof of Stake (dPoS) can lead to validator pools and it is important to understand if these are custodial or non-custodial. Centralized exchanges hold tokens on behalf of their customers and can use this power to vote, and this may be aligned to the agenda of the exchange and not their customers.
- dPoS by definition has token holders who stake, and we have seen that delegators will select the validators who have the highest voting power. If dPoS is in place, then it is important to check the voting power of each validator.
- Even where there are hundreds of nodes and careful research needs doing to understand who controls the nodes.
- Check into the governance mechanisms, who is eligible to vote? is there a written constitution? how does voting work? One token, one vote?. It is worth looking at voter engagement, low turnout may lead to a minority making decisions and this has implications for centralization.
Let’s look at how Tgrade has designed its ecosystem around decentralization. It begins with the legal entity behind the blockchain, there was a conscious decision to form a Dutch Foundation as it is a legal entity in its own right and has no shareholders or beneficiaries.
The PoS has been modified to avoid centralization through several mechanisms:
- Firstly there is a concept of Engagement Rewards which are allocated through a self-sovereign community for active participation in the network. This mechanism incentivizes collaboration rather than cooperation.
- For validators, the engagement rewards are combined with their stake to build voting power. To be eligible for Engagement Rewards a person needs to prove that the Tgrade address they hold is tied to their off-chain identity, and this mechanism ensures that a validator who could run several nodes would only be eligible for one set of Engagement Rewards and thus have a reduction in voting power.
- Tgrade does not have a mechanism for token holders to stake through delegation. This ensures that the validators carry all the risk of the tokens staked, and mitigates the issues around 0% commission attacks or delegators staking with the largest validators.
- There is a sigmoid curve implemented in the distribution mechanism which builds in a disincentive to the accumulation of large stakes by having a mechanism that caps the rewards so that there is no benefit in increasing the stake.
These features combine to ensure active collaboration which in turn makes a strong and secure network, and the layers enforce decentralization.
Why is decentralized important?
Decentralization of a network ensures that the network is robust and secure, through a broad range of independent nodes, and the economic design that incentivizes collaboration. Decentralization is hugely important to secure the assets held on it.
Let’s say to attack a PoS network you need to get your hands on a third of the token supply. It would be an expensive operation, if we took a chain with a market capitalization of $1bn as trying to buy up that amount of tokens will inevitably increase the price of the tokens thus making it more and more expensive.
What if a small group owned over a third of the tokens? If they attacked the chain to reorganize the ledger and steal tokens their holdings would be worth considerably less. The assumption in PoS has always been that the native token which secures the network also provides the safeguards, but in a world of smart contracts where companies issue tokens on the chain with a greater value than the native token, it could present a different threat?
Let’s imagine that there were a couple of hundred sovereign bonds issued on a PoS chain with a value of several trillion euros, the chain attack from the small number of holders with a third of tokens (worth $300m) could provide sufficient temptation? Of course, this is highly improbable, even if they got their hands on the tokenized sovereign bonds what could they do with them if the chain they were launched on had suffered a huge attack? As a thought experiment, it illustrates why decentralized is more robust, and the concentration of power is bad.
The goals of decentralization of networks are in creating a robust, and secure network which is aligned around the common goals.
All is not what it seems in blockchain and while there are headlines around decentralization, it pays to research what the practicalities are. The number of nodes and complex structures can hide an ownership or voting authority issue. Decentralization is not a single thing, rather it is layered, beginning with the design, the safeguards, and economic incentives.