Geographical Decentralisation

Is it actually true that there’s no way to know if a Beacon chain node is hosting a validator or not?

Setting aside that question, wonder if it’s a fair assumption that stake (aka validators) roughly track the distribution shown in this dataset?

Yea, i am thinking the same. I will suss out in few forums that i am in.

1 Like

Hey there,

This is @cortze from MigaLabs, sorry guys for the late reply but the team took some days off and we missed the mails.

Is it actually true that there’s no way to know if a Beacon chain node is hosting a validator or not?

Splitting the Beacon Nodes from Beacon Validators is a security measure that protects validators in the network. The fact that validator duties are public in each slot opens a window of a possible attack if we explicitly know where the validator is hosted.

Setting aside that question, wonder if it’s a fair assumption that stake (aka validators) roughly track the distribution shown in this dataset?

This is a bit more trivial to reply, having a Beacon Node doesn’t really imply hosting a validator. We actually host a few Beacon Nodes just to have direct access to the beacon chain, and Vouch can support multiple Beacon Nodes for a single set of validators. However, this is something worth looking at and we do have it in mind for future studies.

1 Like

I’m fairly sure that Infura, Alchemy etc. will be running a large number of nodes that aren’t validators, I’m sure there are others besides them.

w.r.t geographical distribution of validators, @JustinDrake talked about this on Bankless where he referred to both “jurisdictional diversity” and “geographical diversity”, which is an interesting distinction. If a node is deployed in the AWS datacentre in Beijing, is it in China’s jurisdiction or US jurisdiction or both?

Anyway, I’d like to know what this thread thinks of the idea of self reporting using the grafitti field. This would obviously only be an indicative measure, but perhaps it could be incentivized in some way? Do you think there’s merit in exploring it?

My hunch is node in that AWS dc would be subject to both risks.

Think that’s a great idea, especially for collecting data on at-home stakers (could start with POAPs maybe?). I wonder if the large staking providers (CEXs, professional node runners, etc) might be willing to share and pool this data even without this measure–would obviously require someone to coordinate and run this project / do some degree of outreach to those entities.

Do you think this potentially introduces incremental censorship risk via ability to map validators pubkeys (as well as depositor addresses) to geography?

I don’t think there is a danger of mapping validator pubkeys to to geography if there is sufficiently low granularity (e.g. resolving to country).

I reached out to Lido to ask their advice, and they responded by saying that this is something they’ve put a lot of thought into it, and have a policy around, and is also something that they actively track.

This is a map of where their node opertors are based, from their Q2’22 report:

I think that apart from Lido, it doesn’t really matter where the node operators of other large staking pools reside, as the node operators are bound to the jurisdiction of the company that contracts them.

I still think that it would be useful to cross reference the various datasets with self reported data though. I really feel that have strong jurisdictional diversity is a crucial aspect of a credibly neutral network. I’ll try put some thoughts together on on self-reporting could work in the meantime.

1 Like

100% Agree with your thoughts. mapping at country level is more than enough.

1 Like

I reached to the POAP devs on Discord, and they said that it should be possible to do a POAP drop with validators who self-report using the graffiti field, using the delivery feature.

I found this tool by Raul Jordan at Prysmatic Labs that can help us grab the graffiti.

I looked at some examples of graffiti, and it mostly looks like client versions, e.g.:

“Lighthouse/v3.0.0-18c61a5”, or “teku/v22.8.0”

So I was thinking that we could include a parse-able tag somewhere in the graffiti field using ISO-3166 codes, e.g.: “GEO-BR” for Brazil, “GEO-IE” for Ireland etc.

In terms of how often validators propose blocks, I did some quick sums (assumes all validators have an equal 32 ETH stake):

  • seconds per year: 31,557,600
  • 31,557,600 / 12 secs per block = 2,629,800 blocks per year
  • 2,629,800 / 418,538 active validators = 6.28 blocks per year per validator

Anyway, I’m happy to take this conversation offline and work on this on Discord or whatever. I’ll be at DevCon, and I’m happy to grab a few hours to put together a plan, some docs, and the bones of a webpage - anyone who is interested, just DM me to carry on the conversation.

1 Like

happy to contribute anyway that i can. where would you like to move this convo to ? notion i.e.

I’ll DM you and we can find a suitable platform for collaborating on it. Ideally whatever allows anyone to contribute.

1 Like

better late than never!

1 Like