Really thought provoking article. I like that the framework explicitly makes “decentralization” - a word that gets thrown around a lot - multi-faceted, which allows for nuanced discussions. Vitalik does suggest the placement of each example as rough (and debatable).
I can see how blockchains are, both architecturally decentralized (many nodes hold a copy of the blockchain) and politically decentralized (many people can run their own nodes). Yet, logically centralized (one shared state).
It’s also interesting to note that the difference between Blockchain and BitTorrent is the latter is also logically decentralized. I think Vitalik’s “single monolithic object” vs. “amorphous swarm” analogy is instructive here - in Blockchain, to accurately capture state transition, we need for the state to be “shared” (resembling a single object), while in BitTorrent, once i’ve downloaded a digital file, i just need to make it available as a seed for other downloaders - no one is checking for “shared state”.
To me, this means we can’t just picture “distributed nodes” when we think of blockchains (because that would incorrectly suggest an amorphous swarm); we also need to consider that it is logically centralized.
As the article focuses on “software decentralization”, I found the non-software items more challenging to conceptualize. To his example of the English language shared between Bob and Alice not needed to be related to the English communicated between Charlie and David, I would think that the rules of grammar are “logically” centralized? (Although language is fluid and the rules of grammar are stretched over time, the underlying structure would appear to be centralized?) I’m not so sure about this particular example.
I think the tricky/fascinating thing about decentralization is that it’s a quality that extends beyond software. Decentralization could arguably extend to features of the “network”. I came across a talk by Jameson Lopp that’s influenced my thinking on this:
Source (around 12 minute mark): https://www.youtube.com/watch?v=_IMzSCSeM68
He contends that we must be vigilant in monitoring the level of (de)centralization of various features of a network, including miners, nodes, developers, wallets, exchanges, even non-software features like investors, entrepreneurs and researchers.
Back to Vitalik’s framework; it’s a powerful framework because I think the 3-axes can be extended beyond software. For example, we could talk about political decentralization of blockchain - individuals that control the computers, but also individuals that control payment processors, research funding and agendas, investment capital and so on.