Cooperative consensus vs. competitive consensus: what makes DMDv4 so fast and efficient

In PoW and PoS networks, competition among miners or validators is used as a way to ensure security and stability. However, the same competition also stifles performance and leads to a waste of energy. That’s why DMDv4 uses a cooperative consensus model: a much faster and leaner solution. Read on to find out how competitive and cooperative consensus models differ.

Competitive vs. trustless

In a public blockchain, you can’t have all transaction validators to work together as a team or to be friends with each other. As soon as you give people a chance to collude and manipulate the system, they will do so. Thus, a network where 100% of validators or miners are honest and altruistic and all the users trust them is nothing but a theoretical construct. A blockchain has to be trustless in the sense that nodes and regular users shouldn’t have to trust each other at all.

However, trustless doesn’t mean competitive. There are ways to have nodes cooperate to achieve consensus faster, but without any need for trust. Cooperative consensus is a new model, very different from the traditional Proof-of-Work or Proof-of-Stake.

In particular, DMDv4 uses a variation on one such model, called HBBFT (Honey Badger Byzantine Fault Tolerant consensus), with some additions from dPoS, or delegated Proof-of-Stake. The result is a blockchain that can process up to 400 transactions per second without wasting any electricity, and where transactions become final almost as soon as they are broadcast.

Below, we break down the four key differences between competitive and cooperative networks: transaction processing speed, resistance to attacks, decentralization, and energy consumption. Hopefully, this analysis will make it clear why we chose such a non-standard cooperative solution for DMDv4!

Block time

Shorter block times mean faster transaction processing, but PoW and PoS blockchains can’t afford to cut the block time beyond a certain point. The reasons are two: 1) the stale rate (i.e. the percentage of orphan, or forked, blocks) becomes higher as block time decreases; 2) the risk of centralization grows. (The link between stale blocks and centralization rate is a bit too complex to discuss here, but you can read about it in Vitalik Buterin’s excellent blog post from 2014.).

By contrast, cooperative consensus like HBBFT does not have orphan blocks at all. Instead, validators submit fractional encrypted transactions, and the creation of a new block can start just 1 second after the preceding one, without any increase in the risk of double-spend or centralization.

Verdict: cooperative consensus allows for fast block times without additional centralization risks

Resistance to malicious behaviour and censorship

In the DMDv4 cooperative consensus, validators cannot possibly influence which transactions make it into blocks and which don’t, thanks to encryption. The validator simply doesn’t know what’s in a transaction, because it’s only decrypted after 2/3+1 of validators reach a consensus.

Of course, large competitive blockchains are also very resistant to malicious behaviour like 51% attacks. It’s financially impracticable to attack a network like Bitcoin or Ethereum, but successful attacks on smaller blockchains happen regularly. The latest example is Firo, formerly known as Zcoin.

Verdict: cooperative consensus excludes forks and transaction censorship by validators

Decentralization

This is the only area where competitive consensus models (especially PoS) can have an advantage over the default implementation of a cooperative consensus like HBBFT. In PoS networks, the pool of validators can be huge. For example, as of January 31, 2021, Eth 2.0 beacon chain already had over 75,000 active validators. At the same time, there is a risk that some PoS nodes will become so large as to accumulate both rewards and control over the network (see our recent Medium article for details).

HBBFT limits the number of validators to guarantee lightning-fast consensus: the ‘classic’ version has only 19 validator nodes, and they are always the same. We saw this as a problem, so in DMDv4 we’ve added a clever mechanism of node election, where users stake on candidates. As a result, we can have up to 438 candidates, out of which the system will randomly pick 25 for each epoch (12 hours). We believe that our system is no less decentralized than Eth 2.0 — it’s just different. After all, every user in DMDv4 can influence the validator set.

Verdict: large PoS (competitive) systems are more decentralized by default, but a cooperative network can approach the same level of decentralization through node election mechanics.

Use of resources

It’s a well-known fact that the Bitcoin network consumes more electricity than some smaller countries, such as Czechia. In PoS network, energy consumption per node is low, but if you put together 262,000 validators (as on Eth 2.0), the total consumption becomes very significant.

With cooperative consensus like the one used by DMDv4, you don’t have the energy waste problem. Due to the scarce supply of DMD coins, the maximum number of validators is 438, and the average expected number is between 50 and 75. Between them, they will consume very little energy. As for the staking and voting part, it’s also very energy efficient.

Verdict: the whole DMDv4 blockchain will probably consume less electricity than an apartment block — many orders of magnitude less than a competitive network like Bitcoin.

If you’d like to learn more about our take on the HBBFT consensus, read our White Paper. And if you have any questions for the DMDv4 team, feel free to ask them in the official Telegram chat — we are passionate about our project, so answering questions is always a pleasure!