On-chain Node Beneficiaries 💸 (scalability without sacrificing independence)

Governance NNS Governance
, ,
1

Background

The IC is marketed as operated by “independent parties”, providing the IC’s foundational selling point of “deterministic decentralisation”. Independence is an imperative, not just a nice-to-have. What is meant by independent should be clear, intuitive, and easy to grasp by anyone who may make use of the IC now or in the future. Any formal ties or arrangements between two parties, such as financial or legal commitments, are clear indicators of some level of interdependence or interrelation.

The original idea was that node providers would act as independent parties (1 node from each node provider per subnet). Whether by accident, or intention, node providers are permissibly no longer independent. The best example of this is GeoNodes LLC, George Bassadone and Bianca-Martina Rohner. The former node provider is comprised of the latter two node providers, and so these can only be considered independent if the 3 are collectively recognised as one independent party. Recognition of this has led to the introduction of the concept of Node Providers Clusters, originally proposed on this thread. Clusters then become the “independent parties” that the IC depends on.

The exact definition of what qualifies as a cluster is still to be ironed out, and some are keen to see financial ties between node providers falling outside the scope of what should qualify as a cluster. Discussion related to this aspect can be found on the following thread →

As it stands, the foundation appears to be leaning towards allowing financial ties between node providers without considering them clustered, pointing to the concept of ultimate beneficial owners (UBOs) as a well-defined framework for managing these sorts of relations (or not so well-defined).

Problem Context

Interdependence <> Independence

Allowing for node providers to be financially interdependent/interrelated makes it very hard to argue the case that IC subnets are composed of independent parties. At the very least, it makes the concept a hard sell to those who would come to use and depend on the IC.

Infinite Scalability vs Capital Constraints

At the same time, it’s harder to scale the IC to more nodes if the means of financing more nodes is heavily restricted. Funding/investment, ownership, and potential for control or coercion are concepts that are currently tangled up (or at least not clearly disentangled in a reliable way that NNS participants can clearly see and confidently depend on).

Innovation vs Regulation

Taking a heavy-handed approach regarding restrictions limits potential for innovation in this space, such as the concept of DAO-financed nodes or subsidising nodes via NNS, DFINITY, or SNS grants. Should all recipients of such support be considered to belong to the same cluster? On the other hand, it’s extremely important that the IC doesn’t sacrifice robustness for scalability.

Root of the Problem

Off-chain agreements directly between node providers:

  • limits visibility (hampering decentralisation determinism)
  • challenges the idea of party independence (which is a foundational requirement on the IC)
  • means node providers organise amongst themselves for mutual benefit. There are other activities that fall under this category that would be detrimental to the IC.

A Possible Solution

Nodes, node providers, and node operators are strongly modelled concepts on the IC, and are stored in the NNS registry. For a recap on why the NNS has the concept of node operators (in addition to node providers, see below).

Node operators are represented to reflect the reality that node providers may finance and own the nodes, but that doesn’t mean they’re the ones operating and maintaining those nodes. I would argue that the NNS needs to go one step further, and model the reality that just because a node provider owns the nodes, it doesn’t mean they’re the only one entitled to reward distribution from those nodes.

Relationships between node providers (particularly where money changes hands) should be mediated by the NNS. This decouples the two entities, and removes their need to self-organise for exclusive mutual benefit.

Node Beneficiaries

Rewards are periodically minted to node provider wallets via NNS proposal. Instead of the node providers then dishing those rewards out to their affiliate node provider partners, the NNS should instead recognise those partners as ‘Node Beneficiaries’ and the NNS should divide up and dish out the rewards as necessary (via the same reward proposals that already exist). Importantly, this arrangement is at the continued discretion of the NNS (rather than the parties in question), and is clearly visible to all NNS participants. The potential for leverage between node providers is therefore mitigated.

This also opens up numerous opportunities for the IC:

  • The NNS could act as a match-maker between
    • those with funds who would like to invest in infrastructure growth (but without the hassle of handling node ownership, data centre contracts, etc.)
    • those with technical ability, who would like to own the physical nodes, and who would like to get their hands dirty managing and maintaining nodes (or deciding who to delegate that to)
  • Dynamic Vesting: Percentages could include a vesting schedule (e.g. investor share decays over a number of years)
  • Tokenisation: Beneficiary rights could be tokenised, enabling secondary markets for node revenue streams without touching the node provider
  • DAO Beneficiaries: A DAO canister can appear as a beneficiary, receiving rewards and redistributing internally under its own governance.

There is certainly appetite for these sorts of setups.

If this sort of framework had been in place ahead of the recent node transfers, a large amount of controversy could have been avoided. This model does not prevent node providers and financiers from also striking private side deals, but it removes the necessity to do so.

Ideas have been circulating for a while regarding DAOs that could benefit from node provider rewards (e.g. SNS of node providers, Node Provider DAO, Node Provider Vault). Developers are finally taking bold moves to try and make this a reality. I believe the IC needs something like what’s described above for these sort of arrangements to be feasible.

Please share your thoughts below and help drive this topic forward. There are numerous existing threads that this discussion relates to, in particular:

3 Likes
2

Thank you for sharing your ideas @Lorimer!
To make sure that I understand, we could maybe go through a simple example:

Let’s assume I’m a new node provider with the technical expertise to operate nodes, but without sufficient funds to purchase one.

  • How exactly would the acquisition of my nodes work in this case?
  • Where would the funding come from?
  • And if the NNS were to provide the funds, how would we ensure that I don’t simply take the nodes and disappear?
1 Like
3

Thanks @bjoernek, I’d be interested to know more about how it already works when a large investor (who may already be a node provider) wishes to finance node machines for a new or existing node provider. Isn’t there always the risk that the node providers would ‘simply take the nodes and disappear?’

Either the node provider purchases the nodes and onboards them first, and then receives retrospective compensation, or they onboard one or a few nodes at a time and receive renumeration in stages.

The NNS already has a fund raising mechanism that’s applied to SNS launches. Something similar could work in this case, but with the raised ICP held in escrow. Proof of onboarded nodes could be enforced by a canister that unlocks investment funds for the node provider (or something like that).

Investors would invest at their own risk, which is surely already the case.

In a similar way that prospective SNSs are considered by the community, and many get rejected, the community could take into account qualities that a prospective node provider provably has:

  • Have they just turned up out of the blue claiming an interest in the IC?
  • Do they have a recognisable public presence outside of the IC?
  • Do they already have a presence on the IC?
    • What’s their track record in terms of projects or initiatives they’ve been involved in?
    • Are they already a node provider, and are just looking for help financing more nodes?
    • Are they developers on the IC who would like to help grow out the IC’s infrastructure layer?
    • Are they active governance participants who have demonstrated commitment to the IC?
    • Can they point to verifiable ways in which they are staked in the IC?
  • Can they demonstrate knowledge and technical ability that they’ll need for running these node effectively (without DFINITY holding their hand)?

I understand that what I’m suggesting introduces complexity, but I would argue that it just makes the complexity visible. It’s already hidden complexity that isn’t very well accounted for. If the IC is going to make claims about “Independent Parties”, it should be able to support financing between parties in a way that allows them to continue to be considered ‘independent’.

1 Like
4

Sounds like the nns would act as the middle men between loans and operators.
That might be a good solution but how do people remove bias from the match making?

Would it just be a first come first serve type of situation with rotating lenders, while the node operators stay in line?

1 Like
5

I guess it could be a bit like the neurons fund (in that the raising of funds is decoupled from any specific SNS).

The right to benefit from those funds is separated from the raising of those funds.

6

That sounds like a decent way to remove the cluster issue.

I’m interested in seeing if this moves forward. I think I would only reject a lottery type system. I would really support a rotating lender list with a first come first serve node provider list

1 Like
7

I think neither.

Imo how it should work:

1.when required by target topology, proposal to add nodes in whatever geographical location is appropriate.

  1. After nns approves location any approved nps on waitlist can submit a “bid” to establish nodes at that location. Bid includes documentation of business plan, hardware specs etc proving they have means to meet minimum criteria. And a statement that this is the only source of funding for the nodes (important)

  2. Lowest bidder (in compliance with the specs) gets the funds. (Via another nns vote).

1 Like
8

If an NP submits a bid,

  1. What is the assurance that it can be delivered on the set price and opex that reflect rewards for DAO ?
  2. This entity can simply run away with funds.

Either the capex should be paid over time via a contract where the NP will fund the servers.

2 Likes
9

Great questions. I personally think it would be good for funds to made available retrospectively - pending proof of onboarded and working node(s). This would act both an incentive to help the IC scale and subsidy for the capex.

If opex can’t be covered after accounting for investor shares then that would be a problem. However there are already node providers that have investors with as much as a 24% share. The intention here would be to bring these sorts of arrangements onchain and mediated, rather than to the introduce the concept of financial investors for node providers (it seems to already be a thing).

1 Like
10

Yes, I agree too, then we have the problem where NP / entity has to have these nodes upfront, which affects the scalability problem with the IC node capex problem.

This idea is very interesting. Another question is if DAO pays for these nodes to cover the capex. The ultimate ownership belongs to DAO, yet legally, the bills/invoice is processed/purchased for the NP entity, which then again needs to legally bind the ownership to DAO. Not sure how it would appear in court

Another scenario is that an NP can set up the nodes, get the capex from DAO, then simply unplug and sell out the servers. In a scenario of 42 nodes, that would be a lot of money.

2 Likes
11

Good points, but node providers will need to go through strict KYC and audits. Part of the IC’s deterministic decentralisation concept depends on the idea that node providers can be held legally accountable for any attempts to do damage to the IC (which is why they need to sign a statement to ensure they understand this point - as far as I understand it). If that statement is considered legally binding, there could be another one that ensures that the NP understands they are committing to a minimum timeframe maintaining nodes for the IC (if they’re claiming for a subsidy).

I think there are also other ways of mitigating this risk:

  • A vesting period where the investment is eventually paid off (representing an exit for the investors, and limiting the timeframe where there’s cut and run risk)
  • How much stake does this individual have in the IC (proper ICP stake), and how long are they staked for?

Another way of approaching this is that there’s talk of requiring node providers to build up mandatory ICP stake in the future (to help mitigate other existing risks). The suggestion is that these stakes should be slashable in the event of clear wrong-doing. Perhaps this sort of fund raising / investment approach could be used to help bootstrap those NP stakes.

If node providers are receiving a reasonable portion of their rewards from governance, rather than just infrastructure provision, there would be a number of benefits:

  • Greater resilience to fluctuations in ICP price. This would reduce the pain when the ICP price is steadily falling (with NP rewards lagging behind due to the 30 day moving average). Governance rewards are more stable in this respect.
  • There’s no cut and run risk with staked ICP
  • This would significantly increase trust across the IC as a whole
1 Like