PIP-28: DAN - Distributing the Altruist Network (Community Chains Introduction)

[shane]

Quite a bit to reply to, I’ll try and be as orderly as possible

@msa6867

That being said, in reading your proposal, one would be led to think that this community solution has allowed PNI to shed this portion of its infra costs, or at least the great majority of it. But in reading the comments left by PNI personnel, it seems that PNI plans to maintain substantial in-house altruist infra on top of this community solution.

I think you misunderstand what PNI is currently operating. They have already cut their in-house altruist infra significantly, thanks to CC and others. CC provided PNI with endpoints on February 3rd which enabled them to be able to spin down infra, which the posted about on Feb 13th.

PNI’s cost savings now with DAN is that they don’t need to focus on developing a system that CC already has built. I mentioned what all they would have to address if they were to provide the same level of quality as CC. I don’t know what a theoretical solution from PNI would cost per month, or how long it would take to develop, but CC is an already existing solution which has been done in collaboration with DA, PNI, PNF, POKTScan, Node River, Breezy, qspider, and other ecosystem contributors.

@msa6867

DAN for altruist failover and rare-chain support are different enough issues and same ballpark budgets (ie one is not a small supplemental add compared to other) that the DAO should have the opportunity to vote on them separately

I hear what you are saying, though I disagree since it is all literally the same work. If the work to support a rare chain program was different than the work to support an altruist program, then I would agree, but in this case I feel it’s best to do them together. This proposal was in the works for weeks in collaboration with PNF and PNI, so this paring makes absolutely sense with those close to the rare chain and altruist situations.

@msa6867

If the DAO votes to subsidize rare chains, the subsidy should be to incentivize sufficient on-chain coverage to provide target QoS rather than to build up altruist back up to the unchanged, insufficient on-chain coverage.

Again, I do not see how running a rare chain program or an altruist program is any different… as one would work together with other regardless. Addressing the rare chain need IS addressing the altruists need for rare chains. I do not see the advantage of adding more complexity to two needs that require the same solution.

@msa6867

And I guess it is a little bit about the world, and the ordering of the text of the proposal that may cause a casual reader (and voter!) to miss that this is two separate asks rolled into one.

@msa6867

That being said, you are right, this is an important consideration and could be a reason I would come out in support of a rare-chain proposal… given a chance to be considered in its own right in the manner the subject deserves rather than buried in the DAN proposal.

So far I’ve not heard that anyone is confused by the proposal itself. It is complex, but I don’t believe I’ve burred anything. I appreciate the feedback, but again, I don’t see how rare chain program or an altruist program are better apart.

I appreciate that you say you would support a rare chain program. Instead of asking to separate the proposals, is there a concern you have with CC serving the altruist network and manage a payment model for the node runners under PNF’s oversight?

I would be happy to address any specific concerns right now. I feel addressing specific concerns would be better than coming up with two separate proposals that do the exact same work.

[shane]

@poktblade

This is the concern I have with community chains

@poktblade

The ideal scenario IMO is that this isn’t funded by the DAO given the nature of the product but rather funded by PNI.

I think you misunderstand this proposal as a “funding proposal”, when it’s a service proposal… where we (DA and node runners) get paid for the work we do and value we create for the network. We scoped out this concept in August and began developing in early October. This proposal is not to funding that development… if it were, then it would be a heck of a lot more

This is providing a real-time service where the lion share of the funds go to other node runners, and DA’s fee is minimal, especially considering the service we’ve been providing PNI to contribute to the altruist work. We actually put off launching our public MVP in January because PNI needed a high quality solution for the altruist, and focused on their needs first. We are hoping to launch a public facing version for all node runners soon, now that we’ve addressed the QoS needs of the altruist

@poktblade

This service has the opportunity to drive revenue in so many different directions.

This applies to any service or project in the POKT ecosystem… let along the entire web3 world Your company already provides chain access service to other companies… so should the DAO not work with Nodies since your company monetizes your own infrastructure in ways the DAO can not audit?

While your company monetizes your own infrastructure in the ways you see fit, Decentralized Authority doesn’t own our own node infrastructure and we have multiple parties keeping us accountable. We are strictly a software company. CC Towers are owned by POKT node runners and they are able to verify every relay going through their system. They have full access to ALL logs… because it’s on their system, not owned by us.

Your logic would make it impossible for the DAO to interface with any company (including your own) and that concept should not be applied selectively to CC which right now includes DA, POKTScan, Node River, Breezy, and qspider. We fortunately built our system completely auditable by our node runners, but so many companies in the POKT are not open or distributed… hence why this concept is impossible to implement.

@poktblade

Scenario 2: If DRPC (a load-balancing service amongst node runners) also comes to the forums and proposes something very similar, would we hold the same sentiment?

@poktblade

My team is building middleware software for this - to encourage node sharing. Some of the things that need to be built is bascially a smaller version of infura

@poktblade

V1 solves a lot of the “blockchain” problems we face today, but nothing about the effiency of our how our node runners operate.

I do not see how CC is any different than what you said you were wanting to build. We took a different architectural approach where we weren’t going to be an Infrua like service, but be a distributed service, where we don’t run any infrastructure and node runners themselves own the CC Towers and can audit every relay.

We also built this service to open it up to all node runners, not just select node runners (like other chain access services that exist in the background of the POKT ecosystem). Every product and service we release is freely available to all node runners, and CC is a way we can give any POKT node access to the same opportunities as large providers like Nodies have with their own infrastructure. Even with CC, we are working with every provider that reached out to use to join… we see our “market niche” as node software for any node runner, and that is all we’ve ever released. As @AmishBatman said

@AmishBatman

Our primary business has always been about empowering individual node runners and providing tools for them to do that. That is why we created Node Pilot.

[shane]

@RawthiL

I was thinking that this was not going to be allowed… Although there is no way to control anything in a market like this.

See my comment above @RawthiL. @poktblade was speaking out of conjecture, not facts.

@RawthiL

Is there any restriction for a node runner tapping into CC to consume all of its chains and not deploying any blockchain locally?

This kind of conversation is veering away from what would be the best altruist solution for POKT and it’s customers to discussion chain node pooling.

I would be happy to discuss chain node pooling, how it provides decentralization of POKT node ownership, address chain node over provisioning (which is the largest cost to the POKT ecosystem), and how it incentivizes independent node runners (who took a real hit when providers were capturing market share with closed source geo-meshing, which POKTScan addressed )… but this thread is not the place. PNI was the first to publically discuss it as a ecosystem strategy in October and it is frequently talked about in Node Runners Office Hour, and we’ve been sharing our work on it on since October… so there is a rich history of this convo which we can have elsewhere.

My Ask To Everyone…

Let’s please keep this conversation focused on the proposal and it’s merits.

Let’s also keep our conversation fact based and considerate of all parties

[shane]

@crabman

Question is why should dao fund its development. it would be like funding a competitor because there is nothing stopping you from selling this service to someone else who may not be a pocket node operator.

This is not a funding proposal, it’s a proposal to perform a service to the DAO to address QoS inefficiencies that are effecting POKT customers. While others have submitted proposals for services (like DAO contributions and research), this proposal is about a technical service to address specific needs and involves DA, PNI, PNF, POKTScan, Node River, Breezy, and qspider. Feel free to check out my comment above

If others have a solution that is more distributed, better quality, more accountable, and they feel it will better serve the DAO, they are welcome to do a proposal of their own

[crabman]

Added a ~$20k reimbursement
Added a $8k administration cost
One-Time Reimbursement: ~160k POKT + ~$28k
Monthly: 100k-150k POKT + ~$13k (please see Budget for more details)

“This helps bootstrap the cost of running the CC infrastructure and CC Towers while CC gets off the ground to generate its own revenue”

^ All that sounds like development and management costs, and using DAO funds to get CC to become a viable business.

I am not doubting for a bit that this is not a good service, its just the numbers don’t make sense to me, give me some good solid numbers so I can compare easily. For example, our cost per relay comes out to be x, if you used a competing service it would be y. and so on.

[shane]

@crabman

^ All that sounds like development and management costs

I think where we differ is I wouldn’t call that “development”, but I would call it management cost (which is why this is a service proposal, not a funding proposal).

To be specific about infrastructure costs, more than 1/3 goes to the tower providers since they have been running CC Towers for the altruist network since January. While they have fronted the cost of running these CC Towers, this was a way to alleviate the continued cost until the public service for POKT node runners is off the ground. At this point they have only been running CC Towers for the benefit of the altruist network, so a supplementation makes sense (much like the supplementation for those who would run rare chains).

DA has our own infra costs as well. We’ve spent thousands a month at times just on running out CC testing to ensure capacity and QoS. We are asking for less than $2k to DA to help cover those kinds of costs. Especially with helping launch a rare chain program (which we get no cut of and 100% goes to the node runners), there will likely still be yet more infrastructure required on our side to ensure quality and accountability.

I believe our infra ask is more then understandable, especially considering the amount of work and cost we are taking on.

@crabman

and using DAO funds to get CC to become a viable business.

The POKT ecosystem (through PNI) has already been a customer of CC… I’m just now asking that it be a paying customer, just like the DAO pays for other contributors for their services (similar to community services or research services). Unlike past services though, we have already worked with PNF to provide oversight and have tangible monthly infrastructure costs.

As I mentioned in my proposal

@shane

We expect to have our infrastructure expenses covered through natural CC revenue in a matter of months.

If our providers hit profitability faster than expected, I am happy to discontinue the infra cost. We would still be offering the DAO much more involvement than we would other customers as the DAO has special needs for transparency and oversight compliance with PNF. This is unique compared to any regular CC customer… but the goal is just to supplement some of the infra costs in the very short term.

I open to any feedback about our ask. I did try my best to be as pragmatic as possible.

@crabman

its just the numbers don’t make sense to me, give me some good solid numbers so I can compare easily.

Regarding the relay structure, I would refer to this section as I laid out the exact calculations.

[msa6867]

Can you clarify if the rare-chains portion of the proposal is for specifically and only the three named chains (so that if other chains are found to fall below acceptable QoS levels a new PEP would be needed to fund those new chains), or is for those three chains at the moment plus open-ended addition of other chains in the future according to PNF discretion based on QoS metrics on other chains?

[Breezytm]

My two cents on this proposal for what they are worth. I strongly support the idea behind this proposal. Not only because I am involved but primarily because I do not see a path forward for independent node runners to continue to exist. I am unaware of any cloud providers where one can rent hardware in all three regions to run a Pocket node as well as 15 chain nodes at the current price $10.85 USD a month. Chain pooling gives node runners, independent and providers alike the opportunity to reduce their infra costs. LeanPokt paved the way for this and I believe a solution like CC is the next step forward not only for v0 but in preparation for v1. Finally, all node runners no longer have to sell all their earned tokens at the end of the month instead of re-staking them which from my point of view, from the tokenomic standpoint, this could reduce the tokens in circulation tremendously. This is a net positive.

The sentiment I gather from the responses thus far is that CC can be monetized outside of the Pocket Network while the DAO is funding it in its infancy while it provides support for the Altruist. That is indeed a great and fair point. The solution I propose is the following:

• CC will continue to support the Altruist network and bill the DAO on a monthly basis for the current top 15 performing chains. Subjective to re-evaluation as these chains change based on performance.
• The DAO will offset 100% of the cost of running all remaining chain nodes plus labor.
• DA to agree that these nodes would not be participating in any other activities without a proposal voted on and passed by the DAO.

With these two adjustments, I believe DA can proceed to participate in the open free market if they see fit while the DAO focuses purely on reducing the cost of operation for running the Altruist Network. Do keep in mind, for PNI to deploy and maintain their chain nodes, they need to have one or many employees. As a business, they have to offer these employees both salary and benefits packages. For perspective, community members, such as myself, @qpsider, and others are currently doing so for free.

Let’s not deviate from the bigger picture. Less waste of any form in the network is a net positive for everyone.

[iannn (Ian)]

It IS materially different, because you’re reselling my infra to other node runners and whoever else. What say do I have in that? So basically, no one is allowed to serve altruist traffic unless DA can resell your backends to whoever they want?

[shane]

Hey Ian, working through @Breezytm’s comment at the moment, but wanted to address your comment real quick.

We are not reselling your infra, nor do we have access to it as you are not currently running a CC Tower. I listed our current CC Tower operators above, so I don’t want there to be confusion regarding who the CC Towers are.

I wouldn’t call getting paid by the DAO for a service of value as “reselling”, just like I wouldn’t call the DAO funding a research grant as “reselling words on a piece of paper”. Someone has to do the work to bring everything together and ensure the service is quality (just like researching something for the DAO), so I don’t agree with the concept that DA (or anyone for that matter) should do technical work for the ecosystem without at least minimum compensation.

There is a path to self sustainability for CC through serving POKT node runners, and I’m very transparent about that. But right now, the altruist network is our only client as we have been focused solely on it since January, so I personally believe that is proper for the DAO to be a paying client for the sake of all those involved with CC.

DA’s total cut each month for altruist work will likely be $3k to $4k per month. I personally do not see that as a huge ask, especially since up to half of that helps covers the associated infrastructure costs.

If anyone has specific complaints about the amounts, I’m all ears, but I’d ask that comments on pay or cost be put into proper context. I’m being very transparent and answering every questions I can

Heading back to working on @Breezytm’s comment, as I think it did a good job bringing a lot of comments together. Can’t stay online long, so will likely post tomorrow. Thanks all

[iannn (Ian)]

You would be in the case that the 18 altruist networks I’m currently serving would have to run under CC in order to remain serving altruist. To be clear, you and Fred said connecting my chains to CC is materially no different than connecting to PNI altruist manager. But it IS different, because under CC you would be reselling my infra at your discretion.

[shane]

Addressing Breezy’s Comment

@Breezytm

The sentiment I gather from the responses thus far is that CC can be monetized outside of the Pocket Network while the DAO is funding it in its infancy while it provides support for the Altruist.

Thank-you for this summary. Fortunately we built CC for accountability… literally:

1. Users can’t abuse CC (without being noticed)
2. DA can’t abuse CC Tower Operators (without being noticed)
3. CC Community Tower Operators can’t abuse rare chain operators (without being noticed)
4. PNF controls all DAN funds.

The ONLY WAY for CC to abuse a rare chain would be for DA, CC Tower Operators, and the rare chain operator to create a secret cabal to launch a secret service. Considering the parties involved (DA, POKTScan, Node River, Breezy, and qspider), we can absolutely agree to

@Breezytm

• DA to agree that these nodes would not be participating in any other activities without a proposal voted on and passed by the DAO.

I am happy to require DA and all CC Tower operators to agree to this. CC is designed to only serve POKT, and we are happy to make any formal agreement with PNF to put any doubts to rest.

cc @iannn

A Singular Solution And Proposal

Regarding @msa6867’s point

@msa6867

(2) If the DAO votes to subsidize rare chains, the subsidy should be to incentivize sufficient on-chain coverage to provide target QoS rather than to build up altruist back up to the unchanged, insufficient on-chain coverage.

If the DAO is going to supplement a node, it should be able to benefit on-chain traffic, not just altruist traffic, and be available to ALL POKT node runners equally. CC provides a distributed, neutral, QoS vehicle to do just that for the first time.

The Solana Foundation sponsored the deployment of an infrastructure pool back in 2021. The difference here though is that a POKT solution should only accessible to POKT nodes.

CC provides a permissionless vehicle to do this, as any user can signup their POKT nodes and CC is able to verify if its being used for POKT traffic by monitoring on-chain proofs.

If the rare chain issue is going to be addressed by the DAO, it should addressed for both on-chain traffic and altruist traffic, thus CC makes sense… hence this proposal.

Conclusion

This proposal is predicated on 2 questions:

1. What is best for POKT customers with regards to the altruist network?
2. How do we address rare chains in a fair, transparent, scalable manner, for both on-chain and altruist use?

Regardless which route we take, CC providers as well as some independent providers are currently providing support to ensure the continuity of the network with the assumption of getting their nodes reimbursed with a forward plan. I hope DAN can be that solution.

[AmishBatman (Ryan Burgett)]

@iannn

connecting my chains to CC is materially no different than connecting to PNI altruist manager

To be clear, on a technical level there are significant differences between how Pocket connected chains with their recent altruist manager and how Community Chains connects chains. CC has no central hub that balances traffic and such. People don’t just share endpoints with us and we add them into a mix. We work with providers and they host a CC tower on their own hardware, in their own data center. All traffic is routed via DNS according to location, so if a user (PNI or node runner) makes an RPC request to our network, they get a response directly from the nearest tower. There are no extra hops and no central CC server that manages traffic. We are a distributed network and all requests go directly to our providers.

@iannn

it IS different, because under CC you would be reselling my infra at your discretion.

No, we don’t resell anything from our providers at our own discretion. As Shane said above, all traffic from our users goes directly to our providers. Our providers can verify the numbers themselves and will know if we send them extra traffic.

[RawthiL (Ramiro | POKTscan)]

@shane

Thank-you for this summary. Fortunately we built CC for accountability… literally:

1. Users can’t abuse CC (without being noticed)
2. DA can’t abuse CC Tower Operators (without being noticed)
3. CC Community Tower Operators can’t abuse rare chain operators (without being noticed)

are you going to provide any data endpoints for analyzing the CC traffic? It could be a nice thing to have on POKTscan

[AmishBatman (Ryan Burgett)]

@RawthiL

are you going to provide any data endpoints for analyzing the CC traffic? It could be a nice thing to have on POKTscan

Absolutely. We are making an API endpoint which will spit out altruist stats. Right now, it will show a maximum of 24 hours worth of data with data points at five minute intervals which poktscan can utilize. Each item will include the number of relays by chain over those five minutes and the region. You will be able to pass in an optional time number and if provided it will only send you data points since then (but, still within that max 24 hour range).

[msa6867]

@shane

If the DAO is going to supplement a node, it should be able to benefit on-chain traffic, not just altruist traffic,

Thank you. This answers one of my two questions, and is the answer I was looking for.

Can you please address my one remaining request:

@msa6867

Can you clarify if the rare-chains portion of the proposal is for specifically and only the three named chains (so that if other chains are found to fall below acceptable QoS levels a new PEP would be needed to fund those new chains), or is for those three chains at the moment plus open-ended addition of other chains in the future according to PNF discretion based on QoS metrics on other chains?

[shane]

@msa6867

Can you please address my one remaining request:

Apologies for missing this question. Happy to provide clarification.

Those listed are only three and does not represent all that may require supplementation, however those are the three most expensive and elusive chains due to their resource requirements and service time requirements. This is why I focused on them as an example, as they are the majority of what rare chains would cost. More in depth look at all chains will be begin following this proposal, as mention

@shane

More research will be conducted after DAN becomes an official part of the network

PNF will be providing complete oversight and will be managing all funds. They will pay the rare chain operators directly. The DAO will also have insight into what chains are being suppliment, their cost, and the provider.

PNF is also ensuring that rare chains supplementation is within the nature of this proposal. If after more research it looks like rare chains supplementation will cost more than expected, then PNF can require more DAO approval. However, right now it looks like rare chains will around $10k or so, and all costs will be shared with the DAO.

[shane]

UPDATE: Building off of what @RawthiL suggested about providing an API to POKTScan, CC can also provide a report feed to Discord on the hourly metrics. This means that general data can be accessible directly in Discord, and then POKTScan could have the more detailed view once they integrate the API.

A one line update has been made

[AmishBatman (Ryan Burgett)]

When we say “could” here we mean “will” assuming that Pocket will give us a Discord webhook URL for the hourly stats. It will be the first time the community has real time altruist stats and will allow us to move altruist discussions out to the community where they belong.

[ethen]

I’m in favor of this proposal passing.

From my view, Shane always has the best intentions in mind when it comes to Pocket Network.

Decentralized Authority for the win!