Technical challange to incentivise Relay Node running after Relay Node program ends

Continuing the discussion from Serious lack of transparency and engagement of Algorand foundation:

There was a question asked by @petew.

What will be the long term incentives for relay nodes?

@stephenfoundation answered this as:

The group sometimes referred to as the Relay Node Runners group, came into existence at mainnet launch and should correctly be referred to as Early Backers and Node Runners. This group consists of early investors, commercial entities and non-profits (Universities) who made commitments to support the project and to also run relay nodes to provide the decentralized infrastructure of the blockchain in the early life of the protocol. This group is committed to support the required Relay Node infrastructure of the Algorand blockchain up to 2024. At this point, some of the current participants may decide to discontinue running Relay Nodes. For now, the Foundation has not yet agreed a long term incentive plan to encourage new Relay Node runners to participate, if required, in 2024. The Foundation will engage with the Algorand community, nearer to that date, to evaluate what potential incentive programs are required to onboard new Relay Node Runners. See also technical question below.

and then

As part of our current roadmap, we plan to further the ability to run Relay Nodes on the Algorand Network. One approach being evaluated is to start by using two lists of relays: the current one fully vetted by the Algorand Foundation to keep the network high performance and a second one that is permissionless. Nodes would then connect to relays on both lists allowing best of best world: decentralization + performance. As we move to a permissionless mechanism for Relay Node Running, the Foundation will work with the community to agree an incentive program to support running this infrastructure.

My question is more of technical nature.

Having PPoS, how can we incentivise running Relay Nodes in decentralized fashion? It is pretty easy in the case of other PoS protocols. You stake your coins and choose a validator that takes a % cut from your earnings.

I was trying to think about any way of doing that on PPoS, but nothing comes to my mind. I would even say it is impossible without breaking the basic assumptions of PPoS. I would appreciate it if someone could come with a solution or two to enlighten me.

Thank you.

post by tsachi on May 21, 2021

I don’t know if I have a great answer for that - but here is something that crossed my mind:

• If a relay runner wants to be “paid”, it needs to publish an address (which would get the “payment”).
• Each participating node in the consensus (i.e. that is already voting), would also include in its vote a list of relay account addresses that would be used to send this vote to the network.
• The block proposer for round X would use the cert votes from round X-1 and add these to the block header as “block rewardee”
• A block verifier would make sure the proposed block rewardees match the expected block rewardees.

Note that the above is a bit tricky, since each node might see a different set of cert votes… but I believe that it should be doable.

post by scholtz on May 21, 2021

I believe that the solution is following: Let's allow anybody to run relay node, non-relay node, indexer node, and kmd node...

let's decide that we want to have 100 relay nodes for the network to operate without problems

let's do the competition between the node runners

let's minimize downtimes and request time

simple application can track all nodes, running from different regions to collect stats

at the end of the month / day / hour all node runners will be rewarded from one basket of money dedicated towards them… let's estimate the costs for running the relay node by cutting the budget of relay nodes when there will be a lot of nodes running or increasing it when there are less than 100 nodes…

Competitive solution where, for example, if running one relay node costs $100 / month, and we want 100 relay nodes running it will cost $10,000… incentivizing also other node runners such as indexers or kmd…

In comparison to the current budget of tens of millions $ / month, I believe that the node runners rewards owners already own more than half of all money in the system and there is none chance of change in here.

post by tsachi on May 21, 2021

Having a single entity track and monitor the resources is extremely problematic from a taxation perspective, since it pushes Algorand into being a security. Ideally it would be “the network” which grants these rewards.

post by scholtz on May 21, 2021

Well… they are single entity who “prints money”… who do you think has the key to the main account from which the relay nodes get money? or they are single entity that manages the DNS servers of relay nodes… how does this push Algorand more into being security than it is currently?

This is a list of currently running relay nodes Network Tools: DNS, IP, Email.
Why is every node running under the same top-level domain? There is zero competition in providing better services, and I believe that the node runners may be even so small group of persons that you can count it on the fingers of one hand or max two.

Just admit it that Algorand is totally centralized and stop lying to people.

If I am wrong, please correct me, I see you have flagged Algorand Inc, so perhaps you have more information on this matter.

post by tsachi on May 21, 2021

The list of relays you have provided above is meant to be the bootstrapping list. You can deploy your own relays, and add your own DNS to the config.json, and these nodes would connect to your relays instead of the bootstrapping ones.

Without that list, there would be no starting point for a node to connect to the network.

If, for instance, you would have > 80% of the online money, disconnecting your relays from the bootstrapping relays would allow your network to make progress - so I can’t take your centralization argument seriously. If that’s not truly decentralized, then what is?

(Keep in mind that I never said “fair”. Individuals who control the money, control the network. That’s what the Algorand PPoS is designed around)

post by scholtz on May 21, 2021

Read again my proposal above… if you allow anybody to run a relay node and you incentivize the securing of the network by funding them in a fair way, and the algorithm will put those records to DNS, I call it a more secure and more decentralized network.

Isn't it the main issue of the algo that the 3B was given up to relay node runners? And with the future of governance, they hold more than 50% of all coins and all future decisions?

Also note that the topic is how to incentivize relay node runners… I just bring another solution to the table for consideration. I believe it is much more fair than the current incentives and much more cost-effective for the network to be run securely.

post by krotkiewicz on May 28, 2021

@tsachi thank you for a reply. I am a big fan of your technical expertise and input you provide for Algorand technology stack.

As far as I see, we don’t have any reasonable plan. Going with your suggestion would “centralize” Algorand Relay Nodes - as it is done in other PoS blockchains (they elect validators).

This would be like PoS inside PPoS and this is everything we wanted to avoid in the first place with PPoS, is my understanding right?

post by tsachi on May 28, 2021

I don’t know if the compensation model of PoS (or, more precisely, DPoS) over PPoS for relays is that bad. Depending on its implementation, it could be reasonably balanced. I agree that it would need to be accompanied by a more comprehensive mash-out of the existing relay topology currently in use. Also, there might be some crypto trick where each voter can designate its rewardee, without having the relay know if that’s going to be her or not.

post by krotkiewicz on May 28, 2021

Ok, great, I am getting convinced. We should really have that written down and marketized because the community is seeing this elephant in the room and nobody talks about how to fix that.

In my opinion, this should come as a document before the governance proposal by Silvio and before announcing AVM.

post by Maugli on May 28, 2021

@krotkiewicz what is AVM?

My guess is Automated Valuation Model, a mathematical model for analysis of residential property.

post by Titi on May 30, 2021

@scholtz the 3B is released intermittently based on some factors not all at once. Also 3B is not more than half of 10B.

post by Titi on May 30, 2021

Anyone can run a relay node. He pointed that out. Other nodes could connect to your relay by changing their config file.

post by robert on May 31, 2021

The valid point @scholtz highlighted that only a blessed group of relay nodes are getting rewards for that and there is no competition, no incentives to do it better.

post by robert on May 31, 2021

AVM is Algorand Virtual Machine. Check the announcement.

post by Titi on May 31, 2021

They’re looking to expand that and given ideas about how to. Also, how do you know these runners are not competing? They are competing, it is not like they’re all in the same business or board. Their incentive, I guess, is to have a positive return on investment.

post by scholtz on Jun 9, 2021

Great news… Algorand listens to the community… Algorand Foundation Application Manager - Pilot Program for Algorand Community Relay Node Running.

post by myshkin on Jul 1, 2021

I’m coming in at the end of this convo (very interesting). I’m not a developer or computer science expert, but as someone who wants to see this project flourish, I have to say that the involvement of universities in the relay node architecture was one of the things that most attracted me to Algorand. Even if I don’t know the particulars, I trust that a lot more than some random computer, commercial entity, or libertarian crypto enthusiast running a relay node. I hope the university partnership continues beyond 2024.