ASICs and Decentralization FAQ

[lightclient]

https://download.wpsoftware.net/bitcoin/asic-faq.pdf

Poelstra makes an argument that no consensus algorithm can really be "ASIC resistant", rather some require more capital to design / fabricate. In fact, he argues that PoW operates at Landauer’s lower bound and that as the total hash rate of the universe approaches its theoretical limit, the thermodynamics of heat dissipation forces maximal decentralization.

We are actually better off than the requirement that a single 2N-powered actor has the same advantage as two Npowered actors, because now N is proportional not only to hashing speed but also to heat dissipation requirements, and it is easier for two physically-separated actors to dissipate heat than for just one.

He defines memory-hardness as the amount of memory required by a PoW algorithm to execute. This is often employed as a mechanism to resist ASICs, but since Poelstra argues that consensus algorithms can't be ASIC resistant, all memory-hard algorithms do is weaken the heat dissipation property -- leading to additional centralization.

Additionally, he explores why mining is so centralized now (circa 2015) and how he expects it to change over the next several decades. He believes that electrical costs around the world will become more competitive and lead miners to search for advantages in the form of heat dissipation.