Taulant Ramabaja

February 12, 2015

On centralized vs decentralized ledgers

I can’t count how many times I have heard that the Internet/Bitcoin, etc will “decentralize everything”, that there won’t be any central powers left controlling networks or the means of production. Rarely do people think about decentralization’s effects such as exponentially growing cost of upkeep, and diminishing marginal utility.

“Decentralization is not a goal in and of itself, it is a strategy”.
- Daniel Larimer from Bitshares

On centralized vs decentralized ledgersDecentralized vs Centralized

This is a profound realization, but still insufficient to create a new school of thinking about decentralization in the context of DACs and networks.

To expand our approach to building systems we need to create a new school of thinking about these problems. I propose the following loosely ordered ideas as starting points:

  • All formal systems, including DACs and networks are inherently Axiomatic.
  • Axioms are the set of presupposed inherent truth used as the bedrock of any formal system. In the case of DACs and networks the axioms are not fully known. Therefore, through epistemology we have to think of axioms as known knowns, known unknowns, and unknown knowns (not taking into consideration unknown unknowns).
  • Gödel’s incompleten ess theorems apply to all formal systems, including DACs and networks. This means that systems can either be consistent or complete. Because we are dealing here with computers, obviously the system needs to be consistent, otherwise it could not be computed. Therefore, no system, no matter how intricately designed can be holistic/complete. The best we can do is relative consistency, which also means relative knowledge of the underlying axioms.
  • The end-state of all relatively consistent systems with infinite rounds (i.e. games) leads to a single winner, meaning the game will end in a zero entropy state (i.e. winner take all). This is true only in this very specific setup of relative consistency with potentially changing axioms and potentially infinite rounds.
  • Games (i.e. relatively consistent systems with infinite rounds) can only be played by conscious entities.
  • Only consciousnesses have game theoretical utility values.
  • Consciousness is defined as a thing being differentiated from all else (the universe = U) and other (potential) consciousnesses; Time (entropy) is present; The thing having information about its own state as being separate from U; The thing being capable of computation.
  • Anything, which might also entail an axiom of the system, which does not fulfill all the above definitions of a consciousness is not a consciousness, and as such not a player. Therefore, that thing is just an axiom and can at most be a tool of one or several consciousnesses.
  • Any predetermined executive play between at least 2 players constitutes as contract. Contracts cannot exist without time delay in execution.
  • A trade is an executed contract through which at least 2 players get a higher utility value than before the trade. Other contracts are for semantics’ sake not trade.
  • A market is set of contracts executed among at least 3 players.
  • Decentralization is a strategy for differentiation and specialization in a free market environment.

I believe that the above (most likely incomplete list) is a good starting point to think about decentralized systems and how to approach such system’s design.