Should Platform run on all nodes or should Platform run only on High Performance nodes ?

[TheNetworkIsBiggerThanYou]

thanks for sharing these insights guys. The challenge and magnitude of this effort is coming into focus.

It does seem inappropriate to split up block rewards allocations based on (relatively) short-term technology limitations. Not a good look if the plan would be to fine-tune the initial levels with further changes. Things will quickly get too confusing for anyone trying to follow along. Investors and crypto OG-ers clearly value stability. I'm all for innovation, but there is an optics concern here.

There seems to be too many variables at play here (collateral requirements, block reward allocation, quorum selection criteria, various node types). It quickly gets overwhelming.

Frankly I don't trust a voting system of masternode operators to get it right. Especially without them having the opportunity to see real world data and play around with the tech themselves.

My preference would be to keep it as simple as possible even if that means the product isn't as cheap, fast, or useful as it theoretically can be in the future. Optimization should be added later if necessary. We risk introducing too much complexity that will be very difficult to role back with something more elegant. It is my hope the simplest solution that changes the least amount of things is presented as a viable option to getting a "proof of concept launch". This moving of the goal posts, instead of being directed by DCG leadership (i.e their head on a spike), should come from the MNO community.

Even if this means not enough nodes at first, or poor performance, or disrupted service, I don't see this as a significant down-side especially if it maintains decentralization of the current system and doesn't lock us into any design limitations going forward (like for instance if we rashly choose some combination of these variables that sort of works now for our existing idea of what platform is but not for future ones). For this reason it seems to me the most simple solution would be for core to run nodes financed by the treasury to earn platform fees while they work on optimizing their invention and making it ready for the masses. No new block reward allocation changes. No new collateral requirements. Just fees at first. This will give the devs an opportunity to take a breath and re-assess where they are and will give the network real data to make its future decisions on. As well, it introduces the least amount of changes and risk for investors.

So I support a Proof of Concept Launch to mainnet. Tentatively. Would like to hear from others if this seems viable.

 

[qwizzie]

There is a clear conflict of interest with regards to Dash masternode whales and the upcoming DCG decision proposals.
It is in their own best financial interest (more ROI) or practical interest (easier setup with 10K collateral and less hosting fees) to vote for either the 4K or the 10K decision proposals, while the network is best served keeping itself decentralized.

So it is not only a case of having too many variables at play here, making the decision overwhelming for masternode owners. There is also
a large masternode whale concentration on the Dash network complicating matters and already causing distrust with regards to the outcome of these decision proposals.

So it will not be a case of lets just vote, see what the network thinks and everything will go back to business as usual. These decision proposals will have consequences and possibly have a very negative impact for a very long time. It could very well end up undermining voting participation, when it becomes clear that Dash masternode whales rather vote according their own financial or practical interest, then voting what is best for the network (keeping it decentralized).

 

[TheNetworkIsBiggerThanYou]

There is a clear conflict of interest with regards to Dash masternode whales and the upcoming DCG decision proposals.
It is in their own best financial (more ROI) or practical interest (easier setup with 10K collateral less hosting fees) to vote for either the 4K or the 10K decision proposals, while the network is best served keeping itself decentralized.

So it is not only a case of having too many variables at play here, making the decision overwhelming for masternode owners. There is also
a large masternode whale concentration on the Dash network complicating matters and already causing distrust with regards to the outcome of these decision proposals.

So it will not be a case of lets just vote, see what the network thinks and everything will go back to business as usual. These decision proposals will have consequences and possibly have a very negative impact for a very long time. It could very well end up undermining voting participation, when it becomes clear that Dash masternode whales rather vote according their own financial or practical interest, then voting what is best for the network (keeping it decentralized).

more ROI is not guaranteed. It depends on how big of a block reward is allocated.

It does seem like this change would slow down the rate of even coin distribution though (so less decentralized going forward). A 5 MN owning MNO would be more likely to sell half their coins, than someone running 1 HPMN (4k) and 1MN.

 

[qwizzie]

more ROI is not guaranteed. It depends on how big of a block reward is allocated.

These are the ROI's calculated by DCG research team (ignore the 1K split system, turned out to be incorrectly formulated / impractical / not workable) : https://www.dash.org/forum/threads/...gh-performance-nodes.53374/page-7#post-232371

Just looking at the first one :

At network start (no fees generated) AND Assuming the rewards even out with the market we get the following values..

All nodes run platform: 6.11%
4K split system: 7.05%
10K split system 7.16%

Note : i removed the 1K split system ROI

That is a direct financial interest to Dash Masternode whales, while directly causing centralization on the Dash network.

Of course smaller masternode owners with less than 4000 dash could also be enticed by higher ROI percentages on 4K system / 10K system for normal masternodes (leading to centralization on the Dash network), but it does not also serve a specific practical interest for them (easier setup with their collateral and less hosting fees).

 

[QuantumExplorer]

These are the ROI's calculated by DCG research team (ignore the 1K split system, turned out to be incorrectly formulated / impractical / not workable) : https://www.dash.org/forum/threads/...gh-performance-nodes.53374/page-7#post-232371

Just looking at the first one :



Note : i removed the 1K split system ROI

That is a direct financial interest to Dash Masternode whales, while directly causing centralization on the Dash network.

Of course smaller masternode owners with less than 4000 dash could also be enticed by higher ROI percentages on 4K system / 10K system for normal masternodes (leading to centralization on the Dash network), but it does not also serve a specific practical interest for them (easier setup with their collateral and less hosting fees).

We need to understand what centralization is and why decentralization is important. Decentralization to me is important so the project can remain secure against bad or incompetent actors.

There are a few things to consider: the math shows that under the 4K and 10K HPMN system we are more secure against the top MNO. A 1K Split Platform MN solution doesn't actually make us more decentralized because as you add nodes to the system the % stake of the top whales compared to the rest stays pretty much the same. In fact it even decreases security as the collateral requirement goes down. This is a counterintuitive assessment, but one that is still true. It's because of how probabilities work. The more the size of the pool of nodes the higher the chance of extremity events occurring.

 

[qwizzie]

We need to understand what centralization is and why decentralization is important. Decentralization to me is important so the project can remain secure against bad or incompetent actors.

There are a few things to consider: the math shows that under the 4K and 10K HPMN system we are more secure against the top MNO.

Platform on all nodes solution (no change in collateral)

Security

0.0006% chance to be able to stop a quorum

4K HPM solution (increase in collateral)

Security

0.03% chance to be able to stop a quorum

10K HPM solution (increase in collateral)

Security

0% chance to be able to stop a quorum

Source : https://www.dash.org/forum/threads/...gh-performance-nodes.53374/page-7#post-232366 + below posts there.

That would make only the 10K HPM solution more secure against the top MNO (0% chance of stopping a quorum) compared with all three proposed solutions.
Second is the Platform on all nodes (0.0006% chance to be able to stop a quorum) and third / last is the 4K HPM solution (0.03% chance to be able
to stop a quorum). Correct ?

But what all three solutions have in common, is that these are all very low percentages with regards to risk to security (chance to be able to stop a quorum).
Are these very low percentages really a driving factor in this discussion about setting Platform up on all nodes or setting Platform up through HPM's ?

 

[TheNetworkIsBiggerThanYou]

stopping the quorum has an impact for how long? Does it self-correct or is this the situation in which you'd need the MNOs to reset their nodes?

 

[QuantumExplorer]

Platform on all nodes solution (no change in collateral)

Security

0.0006% chance to be able to stop a quorum

4K HPM solution (increase in collateral)

Security

0.03% chance to be able to stop a quorum

10K HPM solution (increase in collateral)

Security

0% chance to be able to stop a quorum

Source : https://www.dash.org/forum/threads/...gh-performance-nodes.53374/page-7#post-232366 + below posts there.

That would make only the 10K HPM solution more secure against the top MNO (0% chance of stopping a quorum) compared with all three proposed solutions.
Second is the Platform on all nodes (0.0006% chance to be able to stop a quorum) and third / last is the 4K HPM solution (0.03% chance to be able
to stop a quorum). Correct ?

But what all three solutions have in common, is that these are all very low percentages with regards to risk to security (chance to be able to stop a quorum).
Are these very low percentages really a driving factor in this discussion about setting Platform up on all nodes or setting Platform up through HPM's ?

Sorry when I said 1K I meant the 1K split solution you were proposing and talking about before. The safety of "everyone running platform" is indeed the highest for the chance that one MNO can take down the chain. But the "everyone runs platform scenario" has the downsides of forcing people to do something they might not want to do while adding a very high burden to the network in terms of costs to run it. Return rates would go heavily down and platform being successfully started would not be a given.

 

[qwizzie]

Sorry when I said 1K I meant the 1K split solution you were proposing and talking about before. The safety of "everyone running platform" is indeed the highest for the chance that one MNO can take down the chain. But the "everyone runs platform scenario" has the downsides of forcing people to do something they might not want to do while adding a very high burden to the network in terms of costs to run it. Return rates would go heavily down and platform being successfully started would not be a given.

I gave up on the 1K split solution after your post, where you said there was a miscommunication about it. I even stated that i left the 1k split ROI percentage out of it.

The 4k HPM and 10K HPM scenarios have of course the downside of network centralization, deliberately denying smaller masternode owners with less then 4000 dash the opportunity to receive Platform rewards and during software upgrades could even have a higher chance of disruption with number of active Platform nodes dropping to seriously low numbers, then what we are currently experiencing with our v18 update (because the 4K HPM and 10K HPM groups are so much smaller).

With regards to return rates and platform being successfully started :

Return rates will go down on a yearly basis anyways due to Dash emission schedule and Dash yearly cut of 7.1% on its blockrewards. So longterm this is less of an issue. Dash current ROI is 6.73% (?) and that goes down to 6.11% on the Platform on all nodes solution (mayber even a little bit higher if the miner-masternode re-allocation change has not been taking into account). I can live with that. Also most masternode owners are pretty aware for many years now that Dash Platform would raise hardware requirements for masternode owners and would most likely cause higher VPS renting costs as well. We now know those costs are estimated to be $50 for the Platform on all nodes solution, which is good to know.

Platform being successfully started : having a robust PoSe scoring solution for Platform in place would have avoided that. I am not asking for the perfect system here, i am just asking for a system with its essentials remain in tact for Launch. Only now do we find out that there is no robust PoSe scoring solution for Platform, that the lack of having no PoSe scoring solution for Platform creates a risk for Dash Platform launch / future Dash Core upgrades and that there is no intention to create a PoSe scoring solution for Platform before launch. Lets face it, DCG could have subcontracted this to Dash Incubator months ago or ask for additional funding months ago.

It is either poor planning or lack of priority .. or both.

 

[virgile]

Hello everyone!
I am virgile, you might recognize me as that name on DIP 24 and other documents, i'm the resident mathematician or cryptographer of DCG or whatever other title you see fit of giving me. I was told by QuantumExplorer of a discussion happening on this forum.
As you probably know (unless he is that humble), he is currently working an insane amount of hours to fulfill many roles at once and thus struggles to find time to answer to you all with what he considers to be up to his standards. Add to this that i'm in charge of most of the calculations on this side of DCG and you have the following switch in the conversation: I will handle the technical answers, thus please adress me and not QE. I expressedly told him to let it to me, so if QE answers, please let's tell him to get some rest.

First off, I saw a few messages that make me think there might be misunderstandings floating around. So to get on good grounds i will state a few things:

1. To calculate the probability of picking enough byzantine nodes in a validator set to be able to stop it, one must look at the hypergeometric law of related parameters.
2. This probability (let's call it P) is the chance of succeeding in being able to create a stop every time the validator set is refreshed/created. To have the probability of it happening in a period of time where T renewals happen, one must calculate 1 - (1-P)^T.
3. The yield given to only a subset of more potent masternodes supporting platform is an incentive to create more of such nodes. This drives the yield of this node type down and the yield of normal nodes up until an equilibrium is met. Thus having such a separate set of node does not put whales at an economic advantage.
4. It could, however, give whales a stronger control over Platform by driving centralization. The community holds decentralization as upmost importance. We thus decided to give the community a chance to speak up their mind over this choice. We also wanted to showcase how probable a whale takeover would be. We believe it is the best way of acting as what we are: a tool that does what the network wishes. Whale takeover is, i believe, one of the important factors for that specific reason.
5. I have skimmed through the conversation here only very quickly, given my work day has ended a few hours ago and it is soon sleeping time, but I am confident that QE gave you probabilities with a buffer on whales. This means estimating their chances of chain stop on Platform IF they had +20/40% increase of the amount of nodes they owned. I'll verify that statement later on, of course.

Now that we all agree on the basics, we can ensure a proper discourse is at hand and that everyone can form a proper opinion for themselves. I will gladly chat and thus feel free to ask me as much as you wish, state things you are not sure of (so that we can see together if it's correct) or even simply debate. I will try to stay up for a few hours in case any of you wishes to interact now, otherwise i'll regularly check this thread for the following week minimum or until the conversation dies out.

Re Qwizzie : This decision on the number of nodes is, as stated by QE, among other reasons a drive toward a cheaper of usage Platform that has capacity for a higher transaction per second. I have seen stated that this is related to Platform having no Proof of Service. This is as you can see, unrelated. PoSe solves different issues. On that topic, it would take a fair amount of time before being able to implement a proper PoSe and would delay Platform further.

 

[qwizzie]

Re Qwizzie : This decision on the number of nodes is, as stated by QE, among other reasons a drive toward a cheaper of usage Platform that has capacity for a higher transaction per second. I have seen stated that this is related to Platform having no Proof of Service. This is as you can see, unrelated. PoSe solves different issues. On that topic, it would take a fair amount of time before being able to implement a proper PoSe and would delay Platform further.

If the PoSe scoring solution for Platform is unrelated to this discussion, then so is the mentioning of Sam that Platform being successfully started would not be a given for anything but the 4K HPM or the 10K HPM solution. You can't really discuss one, without mentioning the other.

It will be solely a discussion about ROI, Platform Storage Fees and Transaction Per Second, versus sacrifing decentralisation to achieve this (while denying masternode owners with less then 4000 dash the choice / opportunity to receive Platform rewards) and speculating that super low storage fees and higher TPS will somehow translate to a higher adoption rate.

Okay, lets talk about numbers first here : do the ROI numbers that was calculated / estimated for Platform on all nodes, 4K HPM system and 10K HPM system include the miner-masternode re-allocation change that is running till June 2025 ?

See :

https://www.dash.org/forum/threads/...gh-performance-nodes.53374/page-7#post-232371
www.dash.org/forum/threads/updated-mn-miner-reallocation-schedule.51453/

Looks like there is one more change coming up this year in Dash mn-miner reallocation schedule (Nov 2022).

 

[Adding1kDash]

Now that we all agree on the basics, we can ensure a proper discourse is at hand and that everyone can form a proper opinion for themselves. I will gladly chat and thus feel free to ask me as much as you wish, state things you are not sure of (so that we can see together if it's correct) or even simply debate. I will try to stay up for a few hours in case any of you wishes to interact now, otherwise i'll regularly check this thread for the following week minimum or until the conversation dies out.

This precisely the problem I have with the whole discussion, it's not about numbers its about people and its about power. People can't be calulated with a few equation, no matter how fancy are at least for the forseeable future. Centralisation has happend before and it will again. Not saying less than 4000 nodes is perfect, but Dash currently is still decentralised. This can not be said for most other crypto projects. We also know that 101 or less supernode like project have let to centralisation problems ! Once Dash is only 450 or worst 180 HP nodes, the cartel will be formed and it will likely never by undone.

The problem will not be solved by than introducing sharding, because the cartel members will just split there HPnodes themselves. The communication lines will already be in place.

 

[virgile]

Mind that i'll try to answer to every statement seriously, and as such i'll quote you heavily so that you know which points i'm adressing.

If the PoSe scoring solution for Platform is unrelated to this discussion, then so is the mentioning of Sam that Platform being successfully started would not be a given for anything but the 4K HPM or the 10K HPM solution

I said that the introduction of such an idea of 4-10K nodes is not related to PoSe as it's two different topics. It is although, of course, a point of discussion that you can bring up if you want to discuss how to successfully deploy Platform as a project. But that is a broader topic.

(while denying masternode owners with less then 4000 dash the choice / opportunity to receive Platform rewards) ...

Yes unless a system is devised to unify multiple nodes into a beefier node is created, this takes away this opportunity to participate. However, as i stated above, will not create unfairness between both node types as yield equilibrium will be reached.

... and speculating that higher TPS will somehow translate to a higher adoption rate.

This is, as far as i know, not the case. The higher TPS is to be on par and/or do better than possible competitors as to provide an incentive for developers and users to choose Platform instead of other solutions. Will it bring more adoption? perhaps, perhaps not but the way i see it, we do need to create something with clear cut advantages so that adoption starts in the first place, this is a prequisite. Yes marketing will be very important to actually drive adoption, but we still need something substantial behind it.

do the ROI numbers that was calculated for Platform on all nodes, 4K HPM system and 10K HPM system include the miner-masternode re-allocation change that is running till 2025 ? Or is that excluded ?

I saw with Rion and he kindly agreed to answer. QE and I worked on another spreadsheet which makes it hard for me to track what exact parameters QE used. Rion was more involved in this part. Thus he can give an answer which i believe will be more precise.

edit: i will stay up another hour, after that i will answer tomorrow. Statements made while tired are quite unreliable.

 

[virgile]

This precisely the problem I have with the whole discussion, it's not about numbers its about people and its about power. People can't be calulated with a few equation, no matter how fancy are at least for the forseeable future. Centralisation has happend before and it will again. Not saying less than 4000 nodes is perfect, but Dash currently is still decentralised. This can not be said for most other crypto projects. We also know that 101 or less supernode like project have let to centralisation problems ! Once Dash is only 450 or worst 180 HP nodes, the cartel will be formed and it will likely never by undone.

The problem will not be solved by than introducing sharding, because the cartel members will just split there HPnodes themselves. The communication lines will already be in place.

I am unsure of the point you are trying to convey, but blockchain as a field is heavily based on game theory. Such analysis simply follow the path set forth by the whole field previously.

 

[qwizzie]

It could, however, give whales a stronger control over Platform by driving centralization. The community holds decentralization as upmost importance. We thus decided to give the community a chance to speak up their mind over this choice. We also wanted to showcase how probable a whale takeover would be. We believe it is the best way of acting as what we are: a tool that does what the network wishes. Whale takeover is, i believe, one of the important factors for that specific reason.

I have skimmed through the conversation here only very quickly, given my work day has ended a few hours ago and it is soon sleeping time, but I am confident that QE gave you probabilities with a buffer on whales. This means estimating their chances of chain stop on Platform IF they had +20/40% increase of the amount of nodes they owned. I'll verify that statement later on, of course.

Platform on all nodes solution (no change in collateral)

Security

0.0006% chance to be able to stop a quorum

4K HPM solution (increase in collateral)

Security

0.03% chance to be able to stop a quorum

10K HPM solution (increase in collateral)

Security

0% chance to be able to stop a quorum

Source : https://www.dash.org/forum/threads/...gh-performance-nodes.53374/page-7#post-232366

From a technical point of view are these percentage numbers even a point of discussion with regards to Whale takeover ?
Could you please explain why we should be worried about Whale takeover, when the highest risk to security (chance to be able to stop a quorum) is 0.03% or less ? (ironically the highest risk to security seems to be with the 4K HPM solution right now).

I think this is with whales increasing 40% of the amount of nodes they own, while attacking Platform 365 days a year.
I am not sure about the numbers with whales increasing 20% of the amount of nodes they own, while attacking Platform 365 days a year

 

[rion]

do the ROI numbers that was calculated / estimated for Platform on all nodes, 4K HPM system and 10K HPM system include the miner-masternode re-allocation change that is running till June 2025 ?

Yes, and no.

Yes in the sense that the yield calculations use current metrics, which includes the reallocation:
- the miner/masternode allocation right now (39.4% to miners, 50.6% to masternodes, and up to 10% to superblock)
- the number of masternodes right now (3,717 at the time of writing, but it changes in the model over time).

No in the sense that it does not project the allocation over the course of 1 calendar year, because even if we know the allocation over the year, we can't know the number of masternodes over the year, so you'd be mixing knows with unknowns.

(FYI, I refer to it generically as "yield" rather than "ROI". I think it's more accurate.)

 

[qwizzie]

Yes, and no.

Yes in the sense that the yield calculations use current metrics, which includes the reallocation:
- the miner/masternode allocation right now (39.4% to miners, 50.6% to masternodes, and up to 10% to superblock)
- the number of masternodes right now (3,717 at the time of writing, but it changes in the model over time).

No in the sense that it does not project the allocation over the course of 1 calendar year, because even if we know the allocation over the year, we can't know the number of masternodes over the year, so you'd be mixing knows with unknowns.

(FYI, I refer to it generically as "yield" rather than "ROI". I think it's more accurate.)

Pls keep in mind next month there will be a new mn-miner reallocation change.

 

[rion]

Yes, it's in the model as well, and will change automatically.

 

[virgile]

Platform on all nodes solution (no change in collateral)

Security

0.0006% chance to be able to stop a quorum

4K HPM solution (increase in collateral)

Security

0.03% chance to be able to stop a quorum

10K HPM solution (increase in collateral)

Security

0% chance to be able to stop a quorum

Source : https://www.dash.org/forum/threads/...gh-performance-nodes.53374/page-7#post-232366

From a technical point of view are these percentage numbers even a point of discussion with regards to Whale takeover ?
Could you please explain why we should be worry about Whale takeover, when the highest risk to security (chance to be able to stop a quorum) is less then 0.03% ?

Good question. Me and QE are not saying you should or should not worry. in fact, if it was a solution which implied a risk that is too great i would have not let it up to discussion in the first place. We show the probabilities so that you can form an opinion by yourself. We want to show the full picture and give the tools for the poll to be conducted in a proper way.

take 0.03% for example. You want to know the risk of a whale at some point having the ability to block it in a timeframe containing 1000 renewals. You calculate 1- (1-0.0003)^1000. This equals 0.2592 = 25.92%. You must not stop at the 0.03% but calculate over entire timeframes to have a proper idea of the chances of such event happening.

I hope it cleared out why we bring this point up and what to look at. If you have more questions about it do ask.

 

[qwizzie]

Good question. Me and QE are not saying you should or should not worry. in fact, if it was a solution which implied a risk that is too great i would have not let it up to discussion in the first place. We show the probabilities so that you can form an opinion by yourself. We want to show the full picture and give the tools for the poll to be conducted in a proper way.

take 0.03% for example. You want to know the risk of a whale at some point having the ability to block it in a timeframe containing 1000 renewals. You calculate 1- (1-0.0003)^1000. This equals 0.2592 = 25.92%. You must not stop at the 0.03% but calculate over entire timeframes to have a proper idea of the chances of such event happening.

I hope it cleared out why we bring this point up and what to look at. If you have more questions about it do ask.

This means estimating their chances of chain stop on Platform IF they had +20/40% increase of the amount of nodes they owned

What exactly does the +20/40% mean ? I first thought it ment if a whale increased the amount of nodes with either 20% or 40%, but now i am not so sure about that.
If it is true, and earlier calculations were with regards to 40%, then what are the percentage numbers for 20% ?