Discussing EOS' centralization in my previous article sparked some responses from EOS enthousiasts. No one seemed to try to make a case for EOS decentralization though. Instead, the best argument for EOS enthousiasts seems to be pointing at Ethereum and miningpools. So in addition to my previous article, in this article I debunk the arguments for Ethereums "centralization".
"EOS is more decentralized than Ethereum because EOS has 21 blockvalidators, while Ethereum has 3 miningpools."
False. Ethereum has 73 miningpools at the moment. And as I point out in my previous article, the 21 Block Producers in EOS is in reality a fallacy.
"Big miningpools are a sign of centralization and an attack vector. Big pools have too much power + DDoSing those pools could cripple the Ethereum network."
The top 2 Ethereum miningpools provide more than 50% of all hashingpower. (Spark Pool (32%) and Ethermine (21%)) This is true, but to answer the question if that is a risk, we need to understand what miningpools are.
A miningpool is a group of miners that literally pool their resources together so their income of miningreturns is less volatile. So a miningpool, is actually a (sometimes huge) group of miners, that provide hashingpower. These groups can vary over time. A miner can decide to switch pools if other pools seem more profitable. They can also choose to start mining other PoW coins. Miners could also decide to start mining by themselves. This would mean their payout would be less frequent, but when a payout is made, it will be bigger. Whether you single mine, or pool mine, on average payouts should be about the same over a long period of time.
Since one miningpool is actually a collection of independent miners who can switch pools at any time, pools are actually decentralized nodes. And if a pool would decide to use it’s hashing power to run a hard fork and split the chain, they could loose hashing power in an instant. All or most part of the pooled single miners would direct their hashingpower towards other pools or could start mining without the use of a pool. A pool doesn’t have central power of any substance.
⦁ Attack vector:
If a pool would get DDoS-ed (which is not very likely in the first place since professional pools have commercial-grade anti DDoS protection.), then miners who mine through that pool would not be making any mining rewards. Single miners would take action and direct their hashingpower towards other pools or could start mining without the use of a pool. So the decentralized distribution of hashingpower would not be harmed in any serious manner and it would certainly not crash Ethereum.
In addition, it is important to notice is that Ethereum and Bitcoin both have an enormous amount of hashingpower. Halving that hashingpower would still leave more than enough to keep their PoW safe. So even if one or more pools would crash and non of the miners from those pools, (and with them the hashingpower from that pools) would continue to mine elsewhere (which is not a realistic scenario at all), then still it would not crash the network.
So to conclude this argument: miningpools do not make blockchains more centralized and do not form a real threat as an attack vector.
"Ethereum messed up MakerDAO."
Zero bids won auctions. While the Ethereum network was congested, some people managed to abuse the protocol used for loans issued through the Maker Protocol. People place bids of zero value as the only bid for a collateral liquidation auction in a single block. These bids won, were accepted and people walked away with thousands of ETH for free. This obviously resulted in the fact that on the other end of the deal, people lost a ton of value. But the fault was entirely in the design of MakerDAO. Cryptocurrencies are highly volatile and any system can congest. Not taking those factors into account in the their design was an error on the side of MakerDAO: they didn't account for the fact that there could be a time where one single bid ended up as the only bid in one block. That is one unforgivable design error.
Yes Ethereum should scale, that is a fact. But not at the expense of decentralization. For adoption we need chains to be immutable and trust-less. Maybe ETH2.0 will bring a solution. If companies look for centralized blockchains, they will chose blockchains that are run by reputable companies like IBM. In case of errors, they can hold the centralized party accountable. There is no way that anything serious will ever be build on a system that is not immutable and trust-less, while the entities that have actual power in that system are vague entities that either hide behind anonymous addresses or some Cayman Islands based foundations.