Relax. ICP is barely a year old.
1 Like
Markets are currently crashing and people are “losing” a lot of money so tensions are high right now. That’s why we’ve been seeing these meltdowns on the forum.
I tried to explain that in my previous post, but perhaps it needs more background.
Let say your team needs to start a decentralized network like it was never done before. You have some calculations verified on different hardware with different real world emulations. The teams have a feeling how the protocol components will work. The community has some initial docs and papers allowing them to develop and deploy also in a relatively new way.
How do you want to get your platform up? How will the demand relate to the protocol efficiency? How much nodes should be provided at what time? What if we need 1000 nodes a year after launch? Or 2000? What will happen beyond what was emulated before launch? How much time do we need/have for preparation?
You need to do some estimations. You need to implement controls that ensure a stable network with the given resources. This includes human resources. You need to setup rollout plans that can respond to changing factors like delays in the HW supply and service interruptions due to a pandemic, or operational findings because no one did it like that before and other known unknowns.
On the end to need to have a plan that convinces enough node providers to invest in nodes and prepare them to a level that allows any easy onboarding to the network.
“Contributing to the network” thereby only means that the nodes can be added to the initial network at any time. Technically “contributing to the network” means that a node is assigned to a subnet. Spare nodes which are existential for the network are technically not contributing although they have a node ID in the registry. Nodes can have many levels of contribution when building up a platform. You can’t rely on finding the right node provider at the right time and location accepting the right rewards.
The new NP onboarding process only pays rewards for technically contributing nodes. There is no need to prepare full racks around the world. We can onboard what the network needs, or better, what a growth strategy is asking for.
1 Like
What makes you think that? I don’t think the network will grow to a million nodes if they aren’t needed. But you should be prepared for that if that’s your goal, right?
Where do these numbers come from? How much resources are you reserving to ensure a stable network? Are you counting spare nodes or only contributing nodes? How much reserve do you need in a network following a byzantine redundancy that is granulated in subnets? How confident are you with how the protocol efficiency is improving? What do you know about the impact on the network utilisation of upcoming features? Which business plan was changed?
1 Like
The following is the basis for my assertion of less than 10% of compute capacity is being used. Please let me know which assumptions could be invalid / which calcs are wrong/misleading.
- Current ICP provided to NPs: 100,000 /month
- Current cycle burn rate : 5 B/s = 432 T cycles / day = 13000 T cycles / month
- Current ICP Conversation Rate: 9.5 T cycles
- Max ICP Burnt / month FOR cycle consumption: 13000 / 9.5 = 1400 ICP
- Average nodes per subnet does not exceed 40. Therefore a 75% reserve should be adequate.
Comments:
C1. The estimate of 1400 ICP /Month for compute cycles is almost assuredly way too high. But let’s leave it at that.
C2. We are not utilizing more than half of the provisioned compute capacity ( 500 nodes / 1300 nodes).
C3. We should be paying node providers for compute capacity that they provide. Let’s assume that only 25% is effective utilization (accounting for reserve capacity)…i.e. 25000 icp out of 100000 icp.
So 1400 / 25000 = 5.6 % ( less than 10%) of current capacity of 500 nodes is being used AT MAX.
Happy to be challenged or look at alternate methods of calculating.
However if these numbers are even remotely close, what apps are going to cause a 1000 fold increase in capacity spend(millions of nodes) over next two years ; given that this plan was in place pre-launc
It’s understandable. Twitter and ICP trader is probably a better place for venting about price performance.
Average nodes per subnet is way below 40.
Where is the total compute capacity?
Do you understand that we won’t use the full capacity on purpose?
Do understand why we can’t see the current capacity as an over capacity just because it isn’t technically contributing to a subnet?
Beside that you’re now bringing numbers up that need a further understanding e.g. the burn rate only reflects the currently implemented fees.
Our research team is working on a white paper that explains the different price finding models that we could follow in the future. For now I would recommend you to see the over capacity or over rewarding as a subsidy to build up the initial network.
2 Likes
I guess that question is easy to answer if you know what’s currently running on the IC. The vast majority of apps isn’t yet running on the IC. Beside that the number of apps and utilisation per app are two different things with the same result - more load.
I’m open to discuss improvements in the node rewarding but for the initial set of nodes every calculation is correct that allows us to establish the IC with a stable platform. It’s just about adjusting reserves and timeline to gather more data with more projects starting on the IC. Patience pays out here.
This is a good conversation, but it is way off topic for this thread. It would have been nice had this discussion happened in one of the active threads that are currently discussing node provider rewards.
2 Likes
So true! We got quite carried away.
2 Likes
Yes. That is true. We should link the last posts to the threats discussing node provider rewards. And thank you @Luis to carefully add valuable information and reply to all details! Thank you!
2 Likes
so, I am trying to track down how to calculate the daily rewards given over the course of the year for the 8-year stake. For example, say I have 50 staked for 8 years, and every day I vote and receive the rewards. As things stand right now (using the NNS front end) are we expected to leave the rewards untouched (un-merged/spawned) until the end of the year and then calculate it ourselves? or are we supposed to take the initial stake from our account balance sheet, then subtract the merged rewards given at the end of the year to get a number for any forms. This might be more complicated… I did read the medium post… I just am trying to get an idea of what to do/ am responsible for declaring to IRS, or if they currently have it laid out to bypass any taxation until the sale of the entire dissolved neuron over the 8-year time? I am fairly new to this so please bear with me… I just want to maximize my voting rewards while also being mindful of how to properly declare it this coming tax season (2022)