PUP-12: Inflation Stop-Gap Proposal: Double Trouble

And this, my friend, is why I personally have volunteered to spearhead and propose an effort to actively chase down new app dev teams in order to spur the growth in relay demand.

My point is this. Even if rewards are cut in half, won’t we be back in the exact same boat regarding the emission rate of new tokens when the amount of daily relays have doubled!?

What then, half the rewards again, until relays double again?

Seems like a negative feedback loop!?

Giving this some careful thought I realized that this doesn’t represent more than a short term fix at all.

It’s just not possible to cut rewards every single time relay traffic doubles.

What would we possibly do? Lock up billions of tokens in the DAO treasury? Maybe burn 99 of every 100 tokens? This just must have a reasonable, long term solution…

Time locks

A time lock based reward system seems to offer better optics to say the least.

Maybe I’m missing something :thinking:

A lower emission rate with more relays supported by a greater number of projects will serve to make the rewards more resilient. Currently we’re dependent on one or two projects for the relay rewards to support the network. If those projects nuke for any reason we’re up a creek. If we’ve reduced inflation to accommodate those projects then we’re REALLY up a creek.

Also re: a timelock, I think the frequency of the rewards plus the 21 day unstake effectively achieves this without needing to enforce a huge timelock.

But yes, I agree, more projects are needed to support a lower reward multiple.

@michaelaorourke @adam would really love to see numbers around how many nodes are needed to support the number of relays projected for Pocket network over the next year/year and a half.

1 Like

It’s not only possible, it’s necessary. Infura charges $50/month for 200,000 requests per day. That’s 6 million requests per month. On POKT, 6 mil requests at 0.01 POKT/relay and a POKT price of $1.30 costs 200,000 * 30 * 0.01 * $1.30 = $78,000. Take a moment to take that in. Now I understand this is the growth phase. But when POKT reaches maturity and starts burning the app’s stake, it needs to be in a position where it can compete and not cost 1500x more than its competitor. I would argue it needs to get to a sustainable model much more quickly.

So, how do we prefer to get there, a stiff reduction in rewards coupled with a stable/increasing token price that’s attractive to investors, or do we take a dump to $0.001/POKT and keep the same reward level?

2 Likes

Thanks for clearing this up. I appreciate your transparency. I’d like to know why those numbers have never been mentioned prior. Wow. C’est la vie

Someone correct me if I’m wrong, but I believe this has been solved though the PUP- 7 proposal. DApps have a fixed pricing per relay pegged to USD.

1 Like

PUP-7 pegs price of relay for the DApps on the frontend, that’s correct. Over 10 months. In other words, if you’re looking for 1,000,000 requests per day, you stake about 3700 POKT, and at $1.3 per POKT, it costs you about $4,800, over 10-12 months, you’re in line with competitors prices for their mid-range offerings. Not quite, higher, but as PUP-7 puts it, the advantage of a distributed network has to come at a premium.

But if I’m understanding the economics correctly, and please someone correct me if I’m not, nodes get paid per relay, at a rate of 0.01 POKT/relay. So that same DApp which generates 300,000,000 relays over 10 months and “paying” $4,800 for it, will generate 1,000,000(per day) * 30(days) * 10(months) * 0.01 (POKT/relay) * $1.3(USD/POKT) = $3,900,000 worth of revenue for the node runners.

So clearly there’s a big discrepancy between what DApps are paying(or expected to pay) for the service and what nodes are being paid for rendering said service. If you ran a business where you charged the customer 5K and paid your supplier 4 million, you’d be in trouble. Now, I believe this is intentional for the growth phase to attract node runners. But clearly this cannot be a sustainable business model. Inevitably rewards have to be slashed.

4 Likes

I’m loving the debate and interest from so many parties. Everyone is right. And nobody is wrong. This is all our community, so that’s what makes it unique. We just need to reach a consensus on the highest priorities for the protocol as a whole, including all stakeholders - from the supply side to the demand side, capital providers, and community members just here for the culture and the values. Apologises for the diatribe in advance.

Getting down to brass tax, although the timing of this proposal was unfortunate, its importance should not be understated.

Pocket is massively overpaying the supply side. And while the tokenomics are a feature, not a bug; we’ve bootstrapped a global distributed community of nodes into existence. And even if you only joined in the last few months or weeks, you’ve benefited economically, aka the "perpetual fair launch.

WAGMI, truly. Pocket was on its knees for the first few years of existence, and it’s now a global juggernaut about to reach escape velocity.

Consequently, it’s an excellent time to start thinking longer-term about what we’re optimising for.

Is it node growth? If so, what kind of node growth? And where? Or is it something else?

Until v1 comes around, there is no QoS available to reward nodes for the quality of their work, as opposed to the quantity of it.

I think we’re playing a dangerous game if we use Pocket’s economics to reward a neverending arms race amongst node runners. Partly because I actually believe that it will flush out most of the smaller node runners as the technical requirements will just become too high. And it also seems to be a waste of effort and focus. Pocket is more than a protocol with a high APY for participants.

Picking up on a few data points shared by @iaa12 and others:

  • as per PUP-7 - PUP-7: Whole Lotta Requests - Pocket’s centralised competitors get paid - very roughly speaking - $0.00048 to $0.00076 per relay

  • The DAO’s current target cost for the demand side (the “USDRelayTargetRange”) is $0.00036 to $0.00059

  • Currently, the demand side is paying $0. The foundation is paying for all of the relays - via dilution - on all apps’ behalf.

  • Even at the highest end of the USDRelayTargetRange, 10B daily relays would result in around $2.15B in “cost” to the demand side of the protocol.

  • Using the numbers from Token terminal - https://www.tokenterminal.com/terminal/projects/pocket-network- supply side revenue was $3.9m yesterday (Sunday 30 Jan 2022).

  • That’s $1.4B in node revenue annualised per year, without relays or the price of POKT increasing. Of course, this isn’t investment advice, but neither of those things is true when you look further ahead than the next week or two, never mind the next couple of months and longer.

Taking all this together, the recommendations from me(apologies in advance as they respond to numerous concerns / points made in this thread, and aren’t limited to just the protocol’s reward per relay parameter) are:

  1. Start charging larger applications, ie those not on the “free tier” - this will increase the number of staked POKT, and should also create more skin in the game for the protocol’s demand-side as they will now experience greater upside (and downside) from the success of the protocol as a whole, using the price of POKT as a proxy for this

  2. I largely agree with @JackALaing on everything he said regarding the economics: reduce inflation over the next 4 months, but link it to node profitability in USD terms (ie acknowledging how the price of POKT has an impact on profitability). I think that inflation can come down by 50-60% (targeting a rough effective yield of 60-100%). I’m also wary of the circular impact of high incentives arbitrarily increasing node costs due to the need to upgrade infra to keep pace with the rest of the pack, with such costs then used to justify higher inflation rewards. And so on.

  3. we need to ensure that amendments to network-level economic parameters don’t detrimentally impact incentives for longer-tail chains. This means we probably need to review the settlers of new chains economics in tandem with any network-wide changes to the economics. See PIP6.2 for reference - PIP-6.2: Settlers of New Chains

  4. Establish a grants committee and accelerate the rollout of PEP-16 - PEP-16: Universal Contributor Income - we have a fantastic hive mind here. And just like how overpaying for the supply side worked, let’s start overpaying for community contributions. Drive more relays. Increase integrations. Improve documentation. Build more tooling. Create more dank memes. etc etc

8 Likes

Based on all the follow up discussions and clarifications from both the submitter and team, I am now more inclined to support this proposal.

1 Like

i want to add to point 1 you mentioned that a client might not always want to have skin in the game in the company/ protocol providing a service, so another plan should be offered for developers wanting to pay X for Y amount of API requests, like others are offering.

Pardon the list of seemingly random quotes, but I’m hoping to tie this into a comprehensive thought so please bear with me here. First of all, I would like to reiterate the fact that I’ve really enjoyed being part of this community and am thrilled to be joining the conversation here.

One of my biggest concerns here, that I haven’t seen really addressed here is any type of roadmap for an increase in the number of relays that might be expected over the course of the timeframe while we’re looking to decrease rewards for node operators.

If we take a look at relays on a monthly basis, we’re seeing a significant increase, but looking at relays throughout the month of January, they remain fairly steady, ranging anywhere between 255 to 355 million relays per day. That’s great and not insignificant, but as @TimeTravelerBowie pointed out, most of those relays are coming from a single blockchain. If that projects, for any variety of reasons, falls apart, we’re in trouble.

Coinciding with this is a significant increase in new validator nodes on a per monthly basis, starting from November of last year(based on data from c0d3r):

  • November 1 2021: 7463

  • December 1 2021: 13041 (75% increase)

  • January 1 2022: 18176 (40% increase)

  • February 1 2022: 26434 (45% increase)

Are we expecting an increase in the number of relays on a MoM basis to keep pace with node growth? I understand we’re consistently pushing the boundaries on a record number of relates per day, but coinciding with that is a record number of validator nodes coming online per day.

We’re having a conversation around reducing the amount of Pokt earned by validator nodes per day when that number is already moving slightly but steadily down and to the right. I 100% appreciate the need for inflation control, and tying it to relay count is a great idea, but we’re at a point where we’re being told the network is massively over-provisioned and it sounds like we could take on significantly more relays. As another user pointed out, maybe we should be focusing on development efforts to onboard projects to increase the number of relays.

I think there needs to be a conversation happening alongside the conversation around inflation control that is a discussion around how we can effectively onboard new projects that pay for the Pokt network services. We’re treating one side as if it’s still in the growth phase and another side as if it’s in the maturity phase and I think it could be detrimental to the health of the network. We’re trying to balance one side of the scale while not addressing the other.

I understand that we need robust inflation controls and support the conversation around it, but does it make sense to cut the rewards per node as rewards are already sliding down and to the right? I think that is only going to harm node runners who have invested significant capital, time, and effort to help bootstrap the network.

If this doesn’t make sense I apologize, I’m still wrapping my head around everything here

2 Likes

Any argument that we should focus on growing relay count instead of controlling inflation ignores the fact that inflation is directly proportional to relay count, which means growing relays right now (before we’ve implemented an inflation control framework) makes the inflation problem worse. If we grow relays by 2X, we grow inflation by 2X, and the problems being discussed here are 2X worse.

The original WAGMI proposal adjusts the mint rate in proportion to relays so that growing relays won’t lead to more hyperinflation. If we can all agree to implement WAGMI, we can then put this topic behind us and focus on growing relays.

Also, you’re assuming the people debating economic policy are the same people leading business development efforts…

We’re not discussing cutting rewards per node, we’re discussing cutting total rewards. If total rewards (a function of relays) remain static while node count increases, naturally rewards per node will decrease. This is a healthy dynamic because it helps ensure that supply will respond to demand, rather than just growing endlessly. If we try to maintain a minimum reward per node, with no regard for relay count, we will completely unpeg supply from demand and exacerbate the overprovisioning we already see today.

2 Likes

This guy gets it.

We have already bootstrapped the supply-side and we’re now ready for significant spikes in the demand-side.

Now is the time to bring supply/demand into sustainable equilibrium by reducing supply-side payments proportional to relays (WAGMI) and increasing demand-side payments (activating paid tier in the Portal, using Portal revenue to do buybacks/burns, increasing relay count, etc).

3 Likes

Not a negative feedback loop, healthy growth. This is exactly what WAGMI was addressing; we don’t need to vote on a new reduction every time we hit a new relay milestone, WAGMI charts out a reduction at different relay counts.

We already essentially have the dynamic of rewarding stake size. You can opt to consolidate all of your POKT on one node and increase the odds of being selected to produce a block as a validator, since validator selection odds are proportional to stake size, which earns you 1% of a block’s minted POKT. As relays continue to grow, and rewards per node decreases, that 1% will look more and more attractive.

Time lock is a nice idea in theory but it is not possible with Pocket’s architecture, since you’d somehow need a different RelaysToTokensMultiplier parameter per node and/or a different UnstakingTime parameter per node. This would vastly increase the storage requirements of the chain, making it more expensive for everyone to run. The costs would vastly outweigh the benefits. That said, I’m open to exploring whether this is a feature we can develop once v1’s more scalable architecture is implemented, in a later version such as v1.5 or v2.0, as long as everyone maintains realistic expectations that it may not be technically viable or economically desirable.

Does this mean that the team is not going to on-board any more chains/apps until the taper or the bizdev team works independently, the train is going non-stop, and we should just speed up the inflation taming?

The bizdev team is not slowing down their efforts at all. The train is full steam ahead.

I just meant that I see a few people saying “stop talking about inflation when you should be growing adoption”, which ignores the fact that growing adoption makes the inflation problem worse.

2 Likes

Thanks, got it.

Then, if they are going to on-board anyone the size of Harmony soon enough, it would make sense to speed up the taper and personally I would lean to 4-month schedule as opposed to the 6 month one suggested by Michael.

We benefit from the current rewards a lot but everything is good in moderation.

Adam has published an alternative proposal here, an extension of WAGMI that defines a gradual reduction as suggested by many here PUP-13: Initial WAGMI Parameters

well said. i’d like to see some metric targets for any WAGMI proposals. what are the salient numbers - rewards per node? relays per node? dapp staking costs? rewards per dapp staking?

i just don’t get a sense of framework for evaluating inflation good or bad. okay, i see a lot of tokens are being produced. is that bad? is that good? where are we now, where are we headed if no changes are made, and where are we trying to get to?

2 Likes

@JackALaing @lex @iaa12 @o_rourke @Jinx

The point that I find most salient in this thread is the one about massive overpayment to the supply side.

This is 100 percent accurate, we are overpaying the supply side, as so many startups do because it is the hard side of the network. The will, of necessity, end. This is a given.

However, what this point has so far failed to capture is that the overpayment is happening in a commodity, not an asset, and that this commodity is needed for the network to scale.

This is a distinction that is lost on a lot of people who heretofore are used to pricing assets (which is a lot of crypto.) Cash is an asset, homes are assets, stocks are assets. Steel is a commodity, oil is a commodity, grain or lumber are commodities.

Assets are created as a way to store and transfer the outputs of spent energy.

Commodities are the inputs that get consumed during the energy spending process.

Therefore, any economy can be defined as the process of turning commodities into assets.

Q: What happens if an economy runs out of its commodities?

A: It ceases being able to produce assets.

So what does that have to do with POKT?

As POKT is a commodity, the more POKT that exist the more assets (in the form of money) the POKT economy can create. However, the opposite is also true. If we constrain the supply we start to cap the number of assets (stored output) that can possibly be created.

A reduction in inflation would invariably create upward price pressure, however, once all of the available POKT was “spent” in the form of staked nodes, and staked Dapps, the network would no longer be able to grow without either a) re-increasing inflation or b) reducing the number of POKT required to stake a node.

EITHER of these tactics, deployed at a time when the network has hit a mature state, would cause an incredible amount of volatility, which will be much more undesirable in the future than now, potentially unwinding staking businesses, Dapps, and even the network itself.

It is far better to overprovision the POKT network now than risk its underprovision later. I think most of the people on this thread would find that to be an uncontroversial statement. It’s the core assumption in the rewards math in the first place.

It seems to me the debate is by how much we want to overprovision the network, which is a healthy debate and worth having.

But more importantly, I think we need to first agree on what standards we’ll use to measure.

I’d like to suggest that for the sake of the long-term health of the project, we disregard token price and instead focus on market cap.

Looking at Mcap, we can see that inflation hasn’t really impacted things all that much, and the most significant catalyst in terms of negatively affecting price was the exchange listings. Market cap and price have both been relatively stable since listings have enabled more efficient price discovery, with the market cap on a slight trend upward (which is exactly what we want).

IMO if we wanted to stabilize prices for those providing service to the network, the best way to do this would be with a POKT settled futures contract. This is a tried and true way for producers to manage price risk and offload that risk to the speculative market.

Were tokens to be less valuable per token than when some in the network bought, long term that will be positive for the network as it will force or otherwise incentivize large holders to sell, thereby increasing decentralization in the network.

3 Likes