Solution for preventing node provider rewards inflationary death spiral

3 days ago, @Accumulating.icp created a forum thread in which he performed an in depth analysis of the possibility of an $ICP token price spiral taking place as a result of inflationary pressures brought on by node provider rewards payments. That thread can be found here. This analysis has brought attention to just how serious the situation is. As far as I can tell, the Dfinity team has demonstrated very little appetite for providing opinions or commentary of the analysis performed despite several requests to do so. This leads me to believe they are very much aware of this issue and are doing their best to avoid commenting in an attempt to avoid validating the legitimacy of the concerns (which would consequently cause mass panic) while simultaneously not invalidating the concerns (allowing us to continue brainstorming solutions).

Over the last week, we’ve seen the price of $ICP decouple from the its competitor data storage blockchain: Filecoin. It has dropped ranks from 32nd by market cap to 37th by market cap. The market cap of $ICP has gone from $1.5B to $1.3B in the last week and has refused to recover along with the rest of the crypto market. I’m of the opinion that this decoupling is the beginning of an $ICP capital divestment as investors are losing confidence in $ICP’s ability to resolve the issue of inflationary pressures brought on by node providers rewards.

For the last three days, a group consisting of both developers and node providers have held deliberations on protocol changes that can be enacted quickly in order to eliminate the risk of a price spiral taking place. we’ve converged to a proposition that best acknowledges the needs of all parties involved with the parties being the 1.) NNS and 2.) the Node Providers.

The purpose of this thread is to determine the parameters of the proposed changes to the node provider rewards dispersal protocols that are listed below under the title “The Proposal”.

The following proposed changes are intended to be made to the NNS for the purpose of mitigating the risk of the internet computer protocol ecosystem’s token collapsing as a result of inflationary pressures brought on by node provider rewards dispersals:

The Proposal:
1.) A fixed value, denominated in fiat, is assigned as the amount due to be paid to node providers upon each rewards dispersal. Node Provider rewards are dispersed in $ICP (This is currently how node provider rewards are determined and dispersed). The fiat value that node providers are due to be paid is the first parameters that this thread is intended to discuss.

2.) There is a cap placed on the amount of ICP that may be dispersed as payment to node providers. (This cap currently exist within the node provider rewards dispersal protocol, but is set to a rate that is too high to prevent a price spiral). The value of this cap is the second and final parameter that is thread intends to discuss.

3.) Any value that hasn’t been dispersed to node providers (as a result of the cap implemented on ICP node provider rewards dispersals) is tracked as 0% interest debt that is owed to the node providers by the network. This debt is later serviced when the price of ICP has risen enough to permit the NNS to be able to service the debt while staying within the nodes rewards dispersal cap that we decide on.

I will now explain why this proposition was selected as a solution to the issue at hand:

1.) It addresses the issue of run-away inflation as a result of Node Provider payment obligations. It does so by placing a cap on the maximum amount of ICP that is to be dispersed as payment to node providers.

2.) It addresses the issue of neuron stakers being diluted by rewards paid out to node providers. It does so, again, by placing a cap on the maximum amount of ICP that is to be dispersed as payment to node providers.

3.) It transfers a reasonable amount of risk from the NNS over to the individual node providers, subsequently decentralizing risk. It does so by allowing the NNS to receive services from node providers, undisrupted, even when the value of ICP is too low to cover full costs to Node Providers while staying within the rewards cap. The NNS tracks the value owed to Node Providers as accounts payable. This is done at a 0% interest rate and with no hard deadline in which the NNS is obligated to repay node providers. Thus, affording the NNS time to allow the price of ICP to recover.

4.) It takes into account the fixed overhead costs of node providers. It ensures that they are sufficiently compensated in a manner that preserves economic incentive for them to continue to contribute to the network.

5.) It assures node providers have sufficient “skin in the game”. Not by locking up their rewards into a neuron, thereby potentially leaving them unable to pay their overhead costs in the short term. Instead it mandates that node providers invest in the form of providing a 0% interest line of credit to the NNS in times of economic strain. This incentivizes node providers to minimize the amount of sell pressure they contribute to the price of $ICP as they will cease to receive immediate payment for their services if they cause too much price depreciation as a result of over selling their rewards.

This proposal achieves the following:

1.) Decentralization of risk. Currently, risk is centralized to the NNS. If the price of ICP drops below a certain threshold, the result is an inflationary collapse of the entire NNS. If this happens, everyone is screwed, all at once. By placing a cap on inflation generated from node reward dispersals while tracking the balance owed to node providers and later paying that balance down as the price of $ICP rises, node providers are allowed the autonomy that they need in order to balance their own respective balance sheets in accordance to their specific geographical and political environment. On the other hand, if we force node providers to receive their rewards as staked neurons, it raises the already high barrier to entry for new nodes wanting to contribute to the network; anyone wanting to be a node provider would have to have enough money to run an entire data center at a deficit for however long it takes for their first neuron to unlock. This is a huge barrier to entry to becoming a node provider and one that I’m firmly against imposing.

2.) Maintain incentives for node providers to continue to contribute to the network. I’ve considered the perspective of node providers and came to the conclusion that anyone whose willing to buy specific hardware to contribute to the network has at least some level of trust in the network. By implementing a system in which the node providers provide services to the network on credit during times of macroeconomic strain we presume that node providers trust the network enough to issue a 0% interest line a credit. This presumption is not all that farfetched and I assert that it is the presumption that is least likely to negatively impact the incentivization of both current and future node providers.

This proposal does NOT aim to resolve the overarching integral issue of insufficient network activity being the root cause of the NNS’s inability to balance the inflationary pressure brought on by node provider rewards against the deflationary pressures brought on by sufficient network activity. This overarching issue can only be resolved by increased network activity or by an increase in ICP price. Both of which will require time to allow devs to work and to allow macroeconomic conditions to improve. This proposals aims to serve as a set of protocols to be enacted for the purpose of allowing the ecosystem to carry on undisturbed during times in which $ICP price is depressed as a result of the macroeconomic debt/credit cycle.

@Kyle_Langham, your input in this discussion will be much needed as you and @Accumulating.icp have demonstrated yourselves to be the most knowledgable among the community regarding the analytical data that we will be needing to factor in as we determine what parameters are best.

as for @bjoernek , @domwoe and any other developers at dfinity, you all’s participation in this discussion will be vital, as you all are most familiar(or you’re most capable of directing us to the dev that is most familiar) with the portion of the code that would need to be changed in order to implement the parameter changes that we decide on.

as a last word, I would like to remind everyone that an inflationary death spiral could happen suddenly, and could unfold quicker than any of us would have time to respond. There needs to be a sense of urgency regarding this issue. We don’t know how many $ICP whales are out there on the brink of selling their bag which would consequently trigger the beginning of the end.

12 Likes

seems like a decent proposal. however, what about placing some nodes on a “furlough” plan where they don’t need to be online since clearly the network barely is selling any cycles anyway… they can come back online when certain cycles sales KPIs are achieved.

secondly, we don’t know if there are any other bad actor exchanges left. like with FTX where some of the execs had “can’t be liquidated” accounts and could theoretically short the hell out of ICP or any other coin using phantom coins on the exchange’s internal systems…

5 Likes

These brass tacks don’t seem to be garnering the same interest as the last thread, and I asked you to start this fresh one so I’ll humbly kick things off with initial parameters in the hopes that node providers and team members correct me.

I think it sensible to cap NP fiat rewards with current parameter’s, except with an artificial ICP floor (e.g., $2.00, where % inflation sell pressure is est. 15%: Node Provider Inflation Spiral - #41 by Accumulating.icp). The remainder of owed fiat for an ICP below $2 is contributed to a Neuron/wallet at a virtual ICP price of $5. That wallet unlocks if/when ICP reaches $5.

As a simple example, if a node provider is owed $100,000 for the month, and ICP is at $1.50; they are paid 50,000ICP. This shorts them $25,000; and so they get 5,000 extra ICP minted in a locked wallet at the $5 virtual price that’s later unlocked if/when ICP reaches $5.

Just a kickoff suggestion here if I understand you’re solution proposal correctly. I also reserve that if this solution cannot be implemented quickly, it’s better to just lock near dissolved neurons below a certain floor as a ‘band-aid’ to get ‘over the hump’ as described here: Node Provider Inflation Spiral - #117 by evanmcfarland

1 Like

I’m not entirely opposed to this and have considered this, but there is a con (maybe two) that makes this a less attractive option:

1.) Node providers are less incentivized to contribute to a network that could place them on furlough whenever the token price drops. This alone is enough of a disincentivizer for node providers to cause me to shy away from this as an options.

2.) the other potential con to this option is that it would be difficult (potentially impossible) to enforce via the NNS protocol. I’m not sure what steps are involved with furloughing a company from a network, but I get the feeling that somewhere, an intermediary may be required in order to enact the furlough. If that’s the case, it’d move us further from decentralization.

2 Likes

The issue with this is that:

1.) neuron staking isn’t currently coded in such a manner where neurons can unlock once a price threshold is achieved. This isn’t necessarily a huge hurdle to overcome, but it would require more protocol changes. I think the safer thing to do would be to minimize changes to the protocol.

2.) more importantly, this would lead to a substantial amount of sell pressure whenever the price of ICP reaches the threshold where the neurons are unlocked.

2 Likes

the NNS would literally remove their subnet, don’t see how that’s hard to do :slight_smile:

isn’t there some way to see the usage on these nodes? cut out the ones that are in the bottom 50%… or whatever start at the bottom 10%… if price keeps going down, cut another 10%…

i understand this is drastic but companies lay people off all the time when cost cutting is needed during a recession or when there is a shift in the business… this is no different. at the end of the day if the network starts selling more cycles i am sure there will be people lining up to provide nodes… capitalism works really will if we let it do what needs to happen.

i dont see why we need to pussyfoot around. this is a business and these people are at will contractors to the NNS last i checked the NNS is not bound to any long term contracts. regard your pt 1 when there is money providers come out of the wood work don’t worry about that.

does anyone know what is the bare bones number of nodes needed for the IC to stay alive?

2 Likes

Lol. Makes sense.

As to your other point: i understand node providers have to make big investments to start up operations. The difference between a person starting at a company that they can be furloughed from and a node provider deciding to contribute to a network that it could be furloughed from is that the node provider digas to invest tens/hundreds of thousands of dollars up front to begin contributing to the network. The possibility of being furloughed is very much likely to be a big deterrent.

1 Like

again there is no long term contract afaik with any of the subnets. and frankly thats the risk of doing business. data centers buy servers all the time, if they can’t find clients to use them or they lose clients like what happened to them back in 2001 (i’m getting LoudCloud flashbacks here lol) they have to eat the costs… the IC shouldn’t become some kind of welfare system for people making “big investments” when things go south. yes it sucks and no one likes taking out the chainsaw but sometimes it has to be done.

however even with your debt offering you might find that some NPs are unwilling to shoulder this burden and will drop of from the network. my main point is it would be interesting to allow the NNS to decide whether it wants to take on this sort of debt to the NPs or whether it wants to trim some NPs out of the system.

3 Likes

i wonder how many NPs would take this deal, and how many would shut down.

we can’t assume anything, but given that they appear to be selling 50%, it seems that they have about 100% margin built in

1 Like

I shared my two cents on the previous forum thread. I think we need to slow down…

3 Likes

@wpb I do hope you’re right

3 Likes

Given the volatility of crypto markets, and the fact that we are already dangerously close to toeing the line, I think it’s completely reasonable people are proactive in solutions, rather than waiting to get hit in the head before moving.

6 Likes

I disagree. I say we get a stopgap solution in place quickly, revise it after more thought.

5 Likes

Cross posting this forum post with analysis and suggestions for next steps.

6 Likes

Without enough developers and users, inflation will only make things worse.

3 Likes

Thank you very much for your inspiration

Let’s continue to discuss and explore these solutions in more detail or address any specific aspects you have in mind regarding the ICP token economics.

Rebuilt and Reimagined the ICP Customized AI-enabled real-time updates Economics(The economic base determines the superstructure!)Avoiding inflation and death spiral!

2 Likes

@wpb , the negative clickbait headlines have already started. This is one more piece of evidence that suggests a sense of urgency is appropriate. https://youtu.be/HQSAnvCeotM?si=rMWvDdBfk53dtzM5

4 Likes

I was about to say, this certainly does not hepl user adoption, including the lack of a marketing campaign.
Im a big fan and supporter, just a honest comment. The competition is outshining us with marketing and mainstream appearances, and mainstream people will always follow the marketing, and sadly not search some underground project that has the best tech in the world.

2 Likes

To be fair, I do not agree with this statement:

In the very post linked, the third comment is from DFINITY team member talking about DFINITY’s analysis: Node Provider Inflation Spiral - #3 by Kyle_Langham

(not to mention various posts by @bjoernek)

That said, I recognize @bjoernek listened to the intent behind this post and has iterated on how to engage community to address this sentiment, as show by this post: Handling of Node Provider rewards

4 Likes

Fair enough. Perhaps a more appropriate statement would’ve been “dfinity has demonstrated very little appetite for interjecting”

4 Likes