Node Provider Inflation Spiral

Governance
63

Aren’t node provider incentives aligned with a cap on their pay (e.g. 1M ICP/Mo Maximum), at least more than devs getting a price hike. With a dead network, their hyperspecialized hardware just turns into paperweights. Devs aren’t tied to the network in this way and so placing the inflation burden on them seems unwise.

The ICP network, in it’s entire existence, has burned 135K ICP. If your model has it burning 357K ICP/mo, aren’t we talking about a 10X+ increase in hosting cost?

If we can fix this by capping inflationary rewards, can a team member in this thread or other knowledgeable party address what NNS Proposals can and cannot change. Can tokenomics be changed to cap node provider rewards in one way or another? Also, if such a thing jeopardizes network integrity, why not just halt or taper NNS rewards when ICP goes below $3, for example. Neurons and Node Providers are the one’s that should want this the most.

5 Likes
64

This is just poor P&L management from a business point of view!

If INCOME (burn mechanism) has been about 75.000 Icp/year = 325.000$

We can’t have COSTS (node provider rewards + staking) of 75 MILLION $/year (5% inflation at 3$ icp)

It’s basic math… idk why crypto people are so off from a business perspective. They just rely on “price action” and “miraculous growth” but don’t take care of P&L fundamentals!

10 Likes
65

Not an increase in hosting cost. It’d be an increase in network activity.

Currently, there’s only a few applications running on the IC. That means only a few applications are burning cycles. The reason being: to own, operate and maintain an application requires developer experience. The product I’ve built is an entire application that can be owned, operated and maintained without the need for developer experience. This means the number of people who own and operate applications in the IC ecosystem would go from a small number of dev teams, to every single ICP ecosystem participant. Consequently, the amount of cycles consumed would increase proportionately. I really think this is the best possible solution in the mid term and it should be more seriously considered. I’m willing to answer any questions anyone has regarding this.

6 Likes
68

This solution is the only one that addresses the problem at its root. The problem is, there aren’t enough applications running on the Internet computer. This is causing the inflationary pressures of paying node providers to be greater than the deflationary pressures that result from applications running on the IC. The Dfinity foundation has been trying to resolve this issue by attracting more developers- that is a slow process as developers seldom change their web hosting infrastructure.

If growth in the number of applications running on the internet computer is to achieve sustainable rates, we cannot rely on attracting devs in the short/mid term.

The product I’ve built enables consumers to be able to run applications on the Internet Computer, thereby allowing the number of applications hosted on the IC to increase proportionately to the number of consumers within the IC ecosystem as opposed to increasing proportionately to the number of developers in the ecosystem. There are far more consumers within the ecosystem than there are developers.

Again, i have a version one of the white paper available as a PDF in the #white-paper channel of my discord server. I’d upload it here, but this forum doesn’t allow PDFs to be uploaded. I implore you all to read it. I’m here to answer any questions you all may have. I’m fully prepared to defend the product’s business model, incentives, sustainability and any inquiries raised regarding it. Here is the link to my discord server: Personal Dao🥇

5 Likes
69

Burning ICP / Increasing ICP burn rate is not sufficient to offset the NP rewards sell pressure.

Why? Because Burning ICP doesn’t increase demand for ICP.

Burning ICP, and Increasing Demand for ICP are not the same. They are related but burning doesn’t always result in increased demand.

What does burning ICP accomplish? It potentially reduces future sell pressure, and it contracts the supply. Potentially leading to more equilibrium in buy/sell ratio. But it doesn’t necessarily lead to more demand to buy ICP.

Example: If everyone burned their ICP, this would reduce the supply, and it would reduce the amount of ICP available to be sold. So it could reduce total sell pressure. But this would have no impact on the amount of existing NP sell pressure. Only increasing demand for ICP can offset the already existing NP sell pressure.

The theory that burning ICP will increase demand for ICP is flawed. Burning ICP can only indirectly lead to more demand for ICP if the people burning the ICP bought that ICP before burning it. If the people burning the ICP received it as a reward for staking, then burning it doesn’t increase demand for ICP, it only reduces the supply, and potentially reduces the sell pressure.

In order to increase demand for ICP through burning ICP, the ICP must be being burned for a legitimate reason by people who purchased the ICP and then consuming (burning) it to perform an activity on the network.

Conclusion: When users receive ICP for free through staking rewards, and then burn it, it doesn’t lead to increased demand for ICP.

Burning more ICP than is created just leads to a contraction in the supply of ICP. It doesn’t solve the problem of increasing demand. Increasing demand for ICP is the relevant issue to solve the problem of NP reward sell pressure.

Burning ICP and contracting the supply of ICP by slowing the inflation rate doesn’t achieve this.

Even if all other forms of sell pressure were removed through burning ICP, and only NP sell pressure remained, that would not solve the problem of increasing demand or buy pressure.

Increasing burn rate, and thus reducing the supply of ICP might reduce total sell pressure if the ICP being burned would eventually make it’s way to be sold, and thus it would potentially bring the buy/sell ratio into better equilibrium, but it wouldn’t increase buyer demand by itself.

If the burnt ICP was purchased and then burned while being used to perform activity on the network, then that would create demand for ICP. However, increasing the cost of performing various activity on the network would only be feasible up to the extent that it doesn’t reduce demand to perform the activity. Generally, reducing the cost of performing the activity would increase the demand, and raising the cost would reduce the demand.

Therefore I am not sure how increasing the network costs would increase demand, it would most likely reduce demand, and that reduced demand would lead to less ICP buy pressure.

1 Like
70

The product I’ve built allows the application members to be able to stake a shared neuron as a means to paying cycles costs. As their application’s activity increases, so too does their cycles consumption and as a result, they are incentivized to purchase more ICP off of exchanges & stake it in their app’s neuron as their cycles consumption increases.

So the question becomes: how does one incentivize purchasers of the application to onboard users to their application?

The consumer who purchases the application is incentivized to onboard more users to their respective applications as a result of economic incentive. The application has a treasury that is governed by members who are permitted to join by the purchaser of the app. The purchaser of the application is the default admin. Admin receive 1% of all tokens deposited into the apps treasury and 1% of all tokens withdrawn. This is their incentive to onboard new users to their application and grow their treasury, consequently growing their activity, which grows their cycles consumption, which then requires them to purchase more ICP off of exchanges to stake within their neuron.

2 Likes
72

I think it would be reasonable to assume that node providers receive rewards equal to 1:1 of the value of cycles burned by the subnet in which they participate. Initially, however, it can be established that, for example, 20% of voting rewards is a form of funding for node providers.
Additional token minting could only occur to cover the costs of running nodes in subnets that don’t burn cycles (NNS or System subnet).

1 Like
73

With regard to increasing cycle costs. Many of us are making promises to clients and integrators based on current prices. Making this burn rate volatile will greatly decrease the attractiveness of the IC.

9 Likes
74

Ideally node provider hosting costs in data centers could be switched on/off as needed according to network demand, however I don’t think data centers would allow this type of contractual agreement. Has anyone looked into this type of arrangement to assess feasibility?

75

This is not possible with the way the IC is built. Any Node that is responsible for bearing load needs to continue to do so, otherwise Canisters will lose data at scale.

76

@skilesare , I’d like your input on the post that I’m currently replying to as well as the follow up comments I’ve left in response to questions.

2 Likes
77

Hi all, I’ve been following the conversation and love the idea brainstorming going on. I wanted to share a bit more data to help guide the conversations and perhaps identify a timeline of concern.

Below is a chart of FOUND dissolving neurons on the NNS. A neuron is “found” if the id is known by the dashboard. “Found” neurons make up ~27.8M ICP of the 30.1M ICP dissolving in the next 12 months (see dashboard screenshot below) meaning that the bar chart below shows data that is, on average, 93% accurate.

image
image1291×721 46.6 KB

image
image951×474 42.9 KB

For reference, the NNS has typically dissolved between 3 and 5M ICP, so the upcoming 5 months will be slightly below the previous 27 months. There’s a spike to 4.7M in March of next year, followed by a sustained downward trend for the next two years. In the second half of 2025 we see practically no dissolving neurons.

In addition to this analysis, it’s important to account for the potential for more dissolving neurons. Below is a chart of the “found” locked neurons. As would be expected, locked neurons tend to be aggregated around 6-month and 12-month dissolve delays. Should these neuron-owners decide to dissolve, their neurons would be included in the charts above.

image
image1286×727 48.5 KB

I’m adding this data to the conversation because I believe it is relevant to the conversation, however I will refrain from adding my analysis of this data in order to allow white-space for the community to discuss.

Let me know if you have any questions.

8 Likes
78

Thanks for providing this information @Kyle_Langham !

When contrasting it to the previously provided Inflationary Sell Pressure Information, it seems as though the reduction in staking outflow will do a few things:

One of which being greatly reducing the total potential sell pressure.

Alternatively, this drastically up-scales the percentage of inflation sold represented by Node Providers, which degrades the statement that this is a short term non issue as Node Providers don’t represent a considerable amount of inflationary sell pressure.

1 Like
79

The key problem behind this matter is why do we need so MANY NODES and why do we pay them SO MUCH. There shouldn’t be more nodes right node and retribution should be variable depending on network INCOME(burn).
If there was a free market and on-boarding was easier there should be an auction to become a node at the lowest price.

9 Likes
80

Here’s the ultimate solution, the perfect answer. I wish I’d thought of it sooner.

Choose a fixed amount to be minted each month and split it equally between node providers.

If price goes down the highest cost node providers will be the ones forced to shut down.

If price goes up, the increased rewards will incentivize new nodes to come online.

Problem solved. Set it to 200,000 icp per month, split between nodes.

3 Likes
81

Thanks for taking the time to craft a proposed solution! I also originally found this solution lucrative when beginning this dialogue, however it comes with some severe trade offs.

Primarily, we’re talking about an immediate cut of 2/3 in Node Rewards. These funds are used to pay for the nodes themselves in most cases - so there needs to be a degree of profitability otherwise the Nodes will go offline.

In the instance Nodes start to go offline, at scale, the IC is going to run into serious issues.

Each Subnet is comprised of 13 Nodes replicating & verifying data in unison.

If Nodes start to disappear from subnets, the requirement for consensus will drop & collusion becomes dramatically more feasible.

Furthermore, if this happens to a large enough scale, and subnets end up going offline, this leaves us in a scenario in which we lose all of the Canisters & Data that are hosted on that subnet - which is debatably more catastrophic than an inflation spiral.

3 Likes
82

This would require people to be able to run nodes freely. The barrier to entry for running a node on the IC is too big in order to support such a fluid payment system. No data center would commit to purchasing the hardware needed to run an IC node with such risk of having to cease operations whenever the markets are down.

3 Likes
83

The IC needs node hardware that doesn’t require node operators to own entire data centers in order to contribute to the network. This would open the door for the possibility of node operators’ rewards being paid in a fixed number of ICP, as opposed to rewards being denominated in USD and dispersed in ICP. How feasible of a solution is this?

3 Likes
84

Dear DFINITY (cc: the other 20 or so people still reading this forum),

RE: I already said, “I told you so,” but I’m back to tell you one more time.

Okay, apparently, I’m a glutton for punishment, so I’m coming back to weigh in one more time despite the fact that I said this: Absolute Majority Weight Proposal Draft - #66 by FGhostwriting would be my last forum post. Feel free to read it if you want to see my previous prediction of how and why the direction of the network was going from $4.00 to $0.04. It seems to have been completely ignored. Now that we’re under $3.00, I feel a lot like those scientists who warned about global warming decades ago, and now they have to get in their “we told you so’s” because it’s already too late, and they want someone to still be around to hear them say “I was right.”

So now it’s this business about the node providers. And I have to admit, I don’t even remotely have a solution. I hope the Foundation can fix it. After all, they set the blockchain up this way, so they must have had a plan in place in case the token price ever got this low, right? R-right?

I mean, Kyle and DFINITY’s representatives can say all day that we don’t need to worry yet, but we do and they know it and we know it and quite frankly, I think at this point, everyone knows it. We’re under the $3.50 threshold that Kyle warned about back in the day. And let’s be real, we’re not getting back above that number anytime soon. Even FIL is up today, and ICP is still down, breaking the link the two coins have shared for quite some time. The bear market could continue for months, another year, or more. It’s time for the “devs do something” outcry to get louder.

Accumulating has clearly shown the math. We’re already in the downward spiral. Something has to be done now, or it’s over. And to be honest, I’ve gotten to the point where I no longer believe it’s going to happen. Fortunately, I didn’t lock any significant amount of tokens up for eight years (the APR just wasn’t worth it). So, thankfully, I was able to take the huge L for tax purposes and go from having ICP be 95% of my crypto portfolio to a shrinking 20% (the ICP part is mostly in NFTs at this point because I trust pokedstudio with my money more than the Foundation).

The community begged for decentralization. The Foundation said it’s not time yet. Fine. Use your blockchain domination to get us out of this one before it’s actually over.

Signed,

All of us (probably)

18 Likes
85

Thanks for the useful replies, I didn’t realize there was a large barrier to entry. As am Economist, I don’t like barriers to entry.

Doesn’t sound ideal, I don’t know how to fix it.

1 Like