Paqus currently uses a pure Proof-of-Work consensus based on Argon2id, with the design focusing on memory cost rather than raw compute performance.
The idea is to make mining more memory-bound, reducing the advantage of specialized hardware compared to traditional compute-intensive PoW algorithms.
In theory, Argon2 is considered much more ASIC-resistant than algorithms like SHA-256 because increasing memory bandwidth and capacity is significantly more expensive than adding compute units. However, no one can guarantee that this advantage will last forever. If there is enough economic incentive, ASICs could eventually be developed for almost any algorithm.
Our approach is therefore pragmatic: use a well-studied memory-hard PoW today, monitor the ecosystem, and remain open to improvements if the mining landscape changes in the future.
We believe long-term decentralization depends not only on the PoW algorithm itself, but also on fair protocol design and an active community.
Thanks for the detailed explanation.
Using Argon2id as a memory-hard PoW makes sense if the goal is to reduce the immediate advantage of raw compute and specialized hardware.
I agree with your point that no PoW algorithm can guarantee permanent ASIC resistance forever if the economic incentive becomes large enough.
That is why I think the algorithm alone is only one part of the decentralization problem.
The other part is protocol and reward design.
For a community mining layer, the key questions are:
- how to prevent one entity from pretending to be many miners
- how to distinguish active users from passive wallets
- how to make rewards transparent
- how to avoid early extraction before liquidity and utility exist
- how to keep normal users involved after the network grows
So maybe the future is not only pure PoW vs PoS.
Maybe it is also about combining mining with active participation rules and anti-abuse mechanisms.
Argon2id can help reduce hardware concentration at the algorithm level.
LQC-style community mining tries to reduce participation concentration at the reward-layer level.
Both approaches are interesting.
Good luck with Paqus. I will follow how it develops.