Making a New PoW Chain - The Hard Parts
Launching a PoW chain is hard. The trust model is far less forgiving than PoS, where you start with initial token-holders running staking nodes. Making a new type of PoW protocol is even harder. As we get closer to launch, I want to share some of the considerations that go into this process.
Most discussions of launching PoW chains focus on basic game theory or hash power requirements. The reality is more complicated. Most chains are PoS these days. PoW launches are few and far between, mostly because they’re harder to pull off.
Recent history offers sobering lessons: Aleo discarded their pure zkPoW plans after a single entity was able to dominate a testnet through various selfish mining strategies. They ultimately switched to a hybrid proof-of-stake system. SiaCoin nearly imploded trying to make their launch fairer while still establishing a profitable business strategy. Even Bitcoin wasn’t immune – the ASICBoost vulnerability was a protocol flaw in their PoW puzzle that gave Bitmain a structural advantage in mining that persisted for years. This could have had serious ramifications to the chain that, luckily, were overcome at the protocol level.
These problems highlight the requirements for launching a secure PoW chain:
The proof-of-work puzzle must be amortization-resistant. This is exponentially harder for novel PoW designs compared to simple hash-based puzzles. With zkPoW specifically, we need robust guarantees for our whole zkVM – ensuring only one valid computational transcript exists for each puzzle instance. Without this property, clever miners can gain unfair structural advantages that break the protocol’s security assumptions and lead to devastating security losses.
The network needs sufficient honest mining power at genesis to defend against early attacks. The threat isn’t theoretical – botnets performing selfish mining can permanently destabilize young PoW networks. Bitcoin survived its vulnerable early period because Satoshi and a small group of miners provided consistent honest hash power for years. Modern launches face a substantially more hostile environment with much more sophisticated compute ready to be deployed.
The initial mining community must be carefully bootstrapped and incentivized. Early token issuance has to carefully calibrated so that miners can take the risk with a new chain and forego the opportunity cost of directing their compute to other purposes such as training ML models or mining established chains. This trades off directly with long term issuance, which also needs to be sufficient for the chain to retain incentives over time. It can also be difficult to know if solo miners will be competitive with a novel PoW puzzle. This is an issue for zkPoW especially, as generating a ZKP is ~10,000x more computationally expensive than a hash and comes with serious hardware requirements.
Bitcoin’s early mining implementations weren’t optimized. The community optimized the miner over time, and mining grew more competitive. We’re taking a similar approach: no pre-mine and releasing an open-source reference implementation of the mining node. We’ve begun internal work toward software optimization, but will not be able to deploy specialized hardware for some time after launch. This tradeoff mitigates initial botnet risks while ideally distributing the token widely. Overall, the fair-launch approach we’re taking makes our business riskier, but makes the Nockchain mining ecosystem more robust.
The deeper challenge is that zkPoW adds unique complexity to these requirements. Using SNARKs as mining puzzles creates novel attack surfaces around witness generation and proof verification. We need formal guarantees that our zkVM is not only functionally correct, but also satisfies strict cryptographic properties. It’s correct or it isn’t.
Launching a new PoW chain in 2024 demands threading a very specific needle: introducing meaningful innovation, maintaining Bitcoin-grade security assumptions, and getting a large group of miners to try it out.
With zkPoW, we’re playing on hard mode. That’s our moat.