Ethereum is (arguably) doomed to be centralized

I will preface this blog post by saying that my aim here is to set out and list some suggestive, not definitive, evidence I see of increased centralization in the Ethereum ecosystem. It sets out a hunch, not a mathematical proof. If you disagree with my opinion, that’s your prerogative. If you want to convince me otherwise, go dig up the hard data and prove me wrong.

So. Is Ethereum centralized? Isn’t it? The answer, I suspect, depends on who you ask. In the spirit of generating debate, recently I have tweeted about the increasing centralization seen in the Ethereum cryptocurrency ecosystem. See, e.g., the following hard-hitting and dynamic, yet tender and, somehow, ineffably, heartwarming contribution of mine to the corpus of Twitter crypto literature:

Ethereum people were none too pleased:

For those of you who are annoyed with me for this series of tweets, please understand two things.

First, when I start referring to the legendary feats of exploration undertaken by the Marmot Star Empire, that’s generally a good sign that I am pulling your leg.

Second, the blanket assertion that Ethereum is a decentralized system, accepted as gospel by most of the Ethereum ecosystem is, at the very least, arguable. There might have been a time in the past, say 2015-16, where the network could have tolerated the loss of a large number of rank-and-file nodes, selected at random, without much of an impact on the network’s overall functioning. Today that is no longer true.

I would have eventually sat down to write a blog post on the subject but, very fortunately, Twitter user @PaulApivat took the time to read my tweets and summarized them for me in his very considered reply which we should all read. Paul more or less boils down the “Eth is centralized” argument into five pillars:

1. Ethereum is reliant on a handful of private companies to survive.
2. Block reward cuts can be agreed seemingly without objection.
3. Tokens likely remain in few hands, and accordingly so is ecosystem influence.
4. Three entities can collude to reduce mining rewards.
5. Infura dominates the market for node infrastructure available to developers.

Which I would, if starting from scratch, condense down to four:

1. Tokens. The pre-mine looks suspicious as hell. Concentration of large amounts of Ether wealth grants the holder of that wealth outsize influence over the supply of the coins that can be brought to market, including the ability to crash the currency. As put by Muad’dib, “the power to destroy a thing is absolute control over it.”
2. Nodes. The fact that Ethereum has not solved scaling means that centralized service providers, currently Infura, exercise outsize influence over node infrastructure. This is because an archive node now pushes 2 TB in size. The fact that everyone relies on Infura for the system to work, combined with the inability of core devs to find credible scaling solutions, means node counts are falling quickly – and the result is effective centralization in Infura’s hands (which at the end of the day is really just repackaged Azure). (Note, failure to solve scaling is in the interests of the centralizers as it favors them ergo they don’t care about finding a solution. Perhaps this is an accident, perhaps not, but it’s difficult to say for sure from outside.)
3. Clients. There are 13 (or more), but the vast majority of nodes run one of two (Geth or Parity).
4. Too-easy alignment of interests and too-rapid decisionmaking. Major changes like adjustments to mining rewards – changes which would be anathema in other, more longstanding competitors like Bitcoin – are quickly agreed with no objections on the part of major ecosystem players. A lack of public disagreement for changes on that scale makes it likely that those changes are informally agreed before they are formally proposed.

So is Ethereum centralized or decentralized?

I don’t know. But then again, neither do the folks who vociferously assert that Ethereum is the great, decentralized World Computer.

Setting definitional problems to one side (what does “decentralized” actually mean?) I think it is still possible to have a productive discussion about this system based on the commonly-held understandings of “centralized” and “decentralized” among cryptocurrency users and observers.

Earlier in my career, I did a stint in anti-trust litigation. During that time I learned that collusion, where it occurs, is not always apparent to the end-consumer, and is, every single time, informal and unwritten.

The evidence, as I see it, raises red flags that there may indeed be a lot more centralization in the Ethereum ecosystem than anyone realizes.

This will be an unpopular view, especially among Ethereum people, many of whom are my friends. I do not care. Nobody owes the Ethereum ecosystem an obligation to take Ethereum cheerleader-marketers at their word when they tell us that Ethereum is decentralized, or when they say that Ethereum is capable of delivering on promises which assume that Ethereum’s approach to decentralization both works today and is capable of scaling up in the future (see e.g. garbage claims like those made for Plasma, a layer 2 solution endorsed by Vitalik and oft-touted by boosters which claims to enable “billions of transactions per second”).

Truly outrageous claims have been made for Ethereum over the years. The claims are so numerous and diverse that a complete exposition of them does not bear repeating here. But extraordinary claims require extraordinary evidence. And at the moment, even Ethereum’s most basic claim – that it is “decentralized” – should be considered at least somewhat in doubt. Only

• hard-hitting analysis aimed at determining whether collusion has occurred or is occurring in relation to major proposed protocol changes,
• transparency over the extremely mathematically sketchy pre-sale process,
• an honest discussion about the fact that Ethereum can’t handle anything approaching normal daily user traffic for a mediocre web app, and
• more honest discussions about Ethereum’s continuing failure to scale and the likely centralization that is required for Ethereum to continue operating normally under these conditions

will help us get to the truth.

In increasingly greater numbers, reasonable people aren’t buying Ethereum’s lofty pitch. If Ethereum doesn’t like that and is looking for someone to blame, it need only look in the mirror.