Defining (De)Centralization in a Useful Way (The thing you are supposed to be decentralizing is power)
In which I finally explain what I care about.
And what I don’t. A rewrite and reconceptualization of my older “What is Decentralization” and “On Emergent Centralization” fieldnotes.
Decentralization is important because building systems that distribute power is important. Building systems that resist abuses of power is important.
In its simplest definition, decentralization is the degree to which an entity within the system can resist coercion and still function as part of the system. Coercion doesn’t mean force, it means negative incentives to align with an authority.
Note how nothing here explicitly references the distribution of, or the ownership of, entities involved within the system.
Practical centralization is a hard concept to grasp because it isn’t obvious from reading a paper or protocol spec how adoption will play out. However, it is nevertheless something that has to be designed against.
We need to move beyond naive conceptualizations of decentralization (like the percentage of nodes owned by an entity), and instead, holistically, understand how trust and power are given, distributed and interact.
How then do we define centralization in a useful way? And, by useful I mean discriminatory. How do we forge a blade that can be used to cut through bullshit?
Let me first introduce two principles I think are important to consider within this concept.
- Defensive Decentralization: when besieged, a well constructed decentralized system will further decentralize.
- Emergent Centralization: a natural response of systems to non-adversarial conditions e.g. to improve scalability.
These are two sides of the same coin. A system that identifies and attacks emergent centralization will exhibit defensive decentralization. We might say that a well constructed decentralized system will identify & attack emergent centralization.
The questions we must ask of systems must revolve around the recognition and realization of power within the system, and how that system will respond to adversity e.g.
- How is power within the system recognized?
- How is authority within the system distributed?
- How are changes within the system realized?
All of which I think are covered by the single question:
In the face of adversity will the system tend towards distribution of power or consolidation of power?
Decentralization is ultimately about distributing power.
Decentralized systems are systems that rely on the distribution of power to secure the system.
The thing you are supposed to be decentralizing is power
Centralized systems may offer the illusion of distribution of power but ultimately, and especially in times of crisis, the real power must be consolidated in order to ensure the safety and security of the system.
Defining Systems by their response to Adversity
Spam, deliberate attacks, misconfiguration, software bugs, fraud, the pressure of state power; there are many faces of adversity that may impact a system. What matters when considering decentralization is not the nature of adversity, but the response.
If participants in the system my delegate their power to a central authority to respond to adversity, then the system is not decentralized. If participants must trust the same authority, then the system is not decentralized.
If rejection of either of the above result in an inability to interact with the system, then the system is not decentralized.
System designers can arbitrarily limit the impact of any such authority, but putting bounds on an authority does not make the system not-centralized.
Such system nevertheless have uses. Let us not pretend that the authority does not exist though.