Unless I am misunderstanding the data, which is possible: this is disgustingly centralized and easily susceptible to the same problems of cartels, censorship, and manipulation that plague EOS. This is actually MORE centralized than EOS… This is especially disappointing for people like myself who have been trying to become a node provider for almost a year. How many of these 413 are appointed by dfinity? I’m going to guess 100%.
I would also like to know more about this data. @notagain DFINITY has publicly shared their plans to decentralize the boundary and other nodes, allowing anyone to get the hardware, install the software, and create a proposal to the NNS. Are you aware of those plans?
it’s already happening. I think Dfinity has made a gross miscalculation on the importance of decentralization because it was one of the things that didn’t seem to matter as much in the 2017 hype cycle where you had very centralized projects flourishing and seemingly no care given to the decentralization aspect. The market matured and changed drastically and now it is very evident, in the markets and in adoption, just how important decentralization is.
The data in the first post is very concerning. I don’t know why no one but you seems to care about it. To me it is outrageous and should be a primary focus, if not THE primary focus.
That being said, you are aware that DFINITY is actively working toward allowing anyone to run boundary and replica nodes, correct? The current state of decentralization may leave a lot to be desired, but it’s not being ignored.
I agree that perhaps it should be given a much higher level of urgency. We’re somewhat building on sand until the protocol is actually decentralized. It hasn’t even been operating in an environment where Byzantines are possible yet because of how controlled the node operators are. I hope the protocol will hold up.
This is concerning. We need to diversify the distribution of nodes and hosting services. IMO decentralization aspect of nodes does not seem to be a priority for the Foundation or the community right now.
What happened to the Periodic reshuffling of nodes?
Keep in mind that the network was released a year ago, and that DFINITY has already given up a significant amount of control if you look at the voting power of the NNS. Development of the IC protocol is still 99.9% DFINITY, and DFINITY is doing all of the hardware related research.
Let’s not kid ourselves that the IC is decentralized in the same way blockchains that have been around for 8+ years are, but we shouldn’t expect that to happen right away.
How decentralized was BTC or ETH 1 year after inception (honest question)? What are some realistic expectations for the speed of decentralization?
There’s also the hardware requirements and expertise required to set a rig up, not to mention the financial cost and investment risk - it’s not easy.
If you love the IC but want more node decentralization, maybe this is an opportunity for you to get involved in forming a IC community organization that will invest in the hardware to set up and run nodes on the IC? Maybe this is something the community fund looks into as well?
Totally agree with you. it takes time for the network to grow and become stable, when everything is running smoothly and the IC has many dapps that work in reality with a large number of users, the hardware is better researched. then it’s not too late to decentralize nodes, node investors will avoid the risk of hardware investment costs. and the network is verifiable enough to expand simultaneously for investors nodes. I agree with the idea that community funds should also have a part to invest in running nodes. it will ensure that all communities with long-term contributions and interests stick with the IC.
Fair point but we should also keep in mind BTC and ETH weren’t securing hundreds of millions $ worth of assets at that time.
I’m impressed this issue has only recently come up, I’d have expected Dfinity to make sure nodes were spread across multiple providers, they stated there is a huge backlog of nodes waiting to go online. Are these nodes ownee by the same providers mentioned by OP? If that’s not the case why hasn’t Dfinity onboarded some of the nodes on the waiting list to have more variety?
The dashboard UI could also do a better job at displaying this stat, on the main page its stated there are 50~ providers but that number doesn’t mean much if a couple entities own most nodes. It’s be nice to have a UI where the % is shown.
Looking only at the number of nodes per node provider doesn’t give you the full picture. You also have to look at their assignment to the subnets. If a subnet has only 13 nodes then the best you can hope for is that the subnet is composed of 13 nodes from 13 different node providers (who are also distributed well over continents and jurisdictions). This is possible even with the high number of total nodes controlled by few node providers. As a hypothetical example think of 1000 subnets and 13 node providers each having a total of 1000 nodes, one in each subnet. Then 5 (resp. 9) node providers need to collude to break the 1/3 (resp. 2/3) threshold.
In the current situation since we only have 35 subnets the node providers you mentioned will necessarily have 2 of their nodes in the same subnet. This is not ideal and should not happen. But it is clearly an artefact of the IC being in its infancy. Growth in the number of subnets relative to the number of nodes of the largest node provider will eliminate this kind of situation.
The number of nodes per subnet will also increase. The subnets hosting the certain system functionality (NNS, SNS, ledger, cycle minting, etc.) are already larger than 13. We probably want to get to at least the range 30-50.
Virtually all node-related problems are rooted in Dfinity’s projections of demand for the network which were, shall we say, somewhat optimistic. From the Messari report: “The team targets for the network to reach 123 data centers running 4,300 nodes by the end of 2021.”
When the node count reaches 4300, the network will be substantially decentralised.
I’m 100% for more decentralization and post like these help us keep our eye on the ball, but we are also one year in with a mess of technical problems that we’re not anticipated and a bear market. I haven’t ever seen anything implying any plan other than further decentralization and eventually a potential badlands subnet for the adventurous.
multiple instances of collusion between block producers to effectively censor the blockchain, edit wallet balances, manipulate transactions in some nefarious way, thus leading to a screeching halt of all external investment and project interest and development. It still exists but its a ghost ship.
First time I hear this, I doubt it’s true. Afaik a node provider gets back his initial investment in less than 2 years if sells the tokens every month
I don’t understand the rush to make bold assertions that ICP is already worse than EOS as if it’s not publicly known a much larger pool of providers will be onboarded. No blockchain begins its journey fully decentralized. Bitcoin started with Satoshi owning a million of them. I can understand if there was no work being done toward improving infrastructure decentralization.
To dfinity (@diegop@dostro@alexa.smith or whoever needs to hear it):
Please just make sure the minimum amount of decentralization required is there for defi (especially Spinner) and tokenization to take off unimpeded by securities’ or whatever laws it may be. We really, REALLY, R E A L L Y don’t need another Mario.
No matter how many nodes we have as long as they are hosted in data centers they are liable for the content hosted on them, they can either comply or be shut down. Then there is the NNS, which should also be more decentralized and might always take a canister down to prevent legal actions or monetary loss on the provider’s end.