In response to this analysis, Preston has written a thoughtful post elaborating on what he perceives as suggestive evidence that Ethereum is (arguably doomed to be) centralized, which I encourage people to read. I’d like to use this opportunity to examine his reasoning in the hopes of making explicit, certain claims that would be beneficial (and really interesting!) for the community to follow-up on.
Elaborating on the previous analysis, Preston anchors his line of thought around four main themes:
- Too-easy alignment of interests and too-rapid decision making
Claim: Major changes like adjustments to mining rewards are quickly agreed with no objections on the part of major ecosystem players.
I believe I’ve provided data in the previous analysis that indicates that mining rewards adjustments on Ethereum were not quickly agreed upon - that there was debate and discussion for a period of time among a number of parties. Three improvement proposals (EIPs) were put forth and deliberated on.
Dan Finlay provided a helpful flowchart on how the EIP process works for:
Following the process, one can see multiple stakeholders, rich discussion on various social platforms (in addition to previous provided EIPs on github, see Eth Magicians), the All Core Dev calls (#53, all of these are made public); all this discussion before various clients get to choose whether or not to implement the proposed changes, then before miners/users update or don’t.
What the above illustrates is that Ethereum, like Bitcoin, functions under off-chain governance, of which can hardly be accused of being “rapid” nor “easy” in alignment.
Nevertheless, Preston is looking for “hard hitting analysis aimed at determining whether collusion has occurred or is occurring in relation to major proposed protocol changes”. I think the above is sufficient for this particular claim, but one’s mileage may vary. On a side note, Preston’s draws on his experience doing anti-trust litigation in making this claim; I would be very interested to read about methods for detecting collusion (if any) that could be brought to bare.
I do think it would be helpful for “major ecosystem players” that’s implied in the claim to be made explicit. That would allow for follow-up analysis for verification.
Claim (regarding clients): There are 13 (or more), but the vast majority of nodes run one of two (Geth or Parity). (The implication is that when there a majority of nodes run on two of 13 (or more) clients, there is a degree of centralization.)
I can verify this claim. Preston’s point is valid and well noted here. Geth is the dominant client, followed by Parity-Ethereum and Parity.
This is something the Ethereum community is aware of and even then, given the 10 or more additional clients, it would be worthwhile for the wider community to monitor adoption/usage rates of other clients as a measure of relative centralization-decentralization moving forward.
Claims regarding Nodes:
It was acknowledge in the previous analysis that the point about Infura’s dominance in the market is valid. And that several teams are working on alternative to Infura. Not to mention the argument that “a strong network of thin and light clients” could be an effective alternative. (I’m aware some would contest this last point, most notable argument put forth here. It’s a meaty post, so I’ll save it for separate analysis).
Nevertheless, Preston makes several sub-claims which may be interesting for the community to dig into including:
Larger claim: 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.
Sub-claim: Everyone relies on Infura for the system to work.
Some interesting evidence would be node distribution (actual counts) among three separate types: full nodes; light nodes; archive nodes.
Sub-claim: Node counts are falling quickly.
Assuming he’s referencing full nodes, evidence to verify/falsify this claim would include number of nodes over time (say between 2015 - 2019).
Final theme: Tokens
Claim: 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.
This is an interesting claim. The context for this is his suspicion around Ethereum’s pre-mine process. The argument is that a pre-mine leads to a concentration of large amounts of Ether in a few hands. The counter-argument is that the team behind Ethereum worked hard for no pay for a long time to launch the protocol and they knew further research, development would be needed to bring the vision to fruition and a pre-mine rewards the biggest risk takers.
In the previous analysis, I made a case that previously concentrated wealth has been meaningfully distributed in the form of research grants to teams doing real work in the ecosystem.
Even so, it would be interesting to see evidence of irregular price movement for ETH to verify/falsify the idea that “influence over the supply of coins…including the ability to crash the currency”. To my knowledge, the bear market has affected ETH as much as BTC and every other coin other there. I haven’t seen evidence of price tampering.
As a thought experiment, let’s suppose that a large concentration of Ether fell to a sub-group and that group wished to exercise outsized influence.
Some would argue that part of the motivation to move to proof-of-stake is to counteract any “outsized influence” a sub-group (i.e., cartel) could have based on their coin holdings. Strategies have been proposed to discourage centralization, with specific incentives to discourage joining large pools, discourage putting nodes in the largest virtual private server provider (i.e., Infura) and encourage the use of secondary software implementations (i.e., the other 10 aside from Parity and Geth). Source: https://github.com/ethereum/wiki/wiki/Proof-of-Stake-FAQs#are-there-economic-ways-to-discourage-centralization
Of course, all of the above require empirical data, which will not be clear until the system is actually running. To that end, validator behavior and the incentive structures that are implemented would be important to monitor when the move to PoS is completed.
All in all, I think Preston’s skepticism is healthy for the Ethereum ecosystem. I do not think the centralization-decentralization debate is meant to be “settled”, but rather indicators of each to be continually monitored moving forward. Addressing the claims made explicit and the suggested evidence would be a step in that direction.