• Forum has been upgraded, all links, images, etc are as they were. Please see Official Announcements for more information

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

@qwizzie thanks for the continued posts. The picture is no clearer for me now after reading through this discussion. The 1k option now doesn’t work, but keeping the collateral at 1k is one of the options. Head scratcher!
 
@qwizzie thanks for the continued posts. The picture is no clearer for me now after reading through this discussion. The 1k option now doesn’t work, but keeping the collateral at 1k is one of the options. Head scratcher!

Yeah, i think the 1K split system option would perhaps involve too much change to get it working. It would most likely need more masternodes to keep InstantSend working optimal over both assets (IS optimal for 1K split system means 2x 1920 nodes, while we currently only have 3701 active masternodes on the network, so too low to achieve that), which possibly means changing the 1K collateral to the down side to stimulate the setup of more masternodes. This is a change with a number of effects : it would affect VPS hosting fees, time interval between MN payments, number of MN payments for masternode owners. And the rewards from Platform and Core between both assets would also need to be structured in such a way, that it balance rewards out evenly over 1K Platform nodes and 1K normal nodes.

I am still confused how DCG research team came up with those ROI percentages for normal nodes and Platform nodes on the 1K split system. As that seems to have worked out how to balance the rewards part of the 1K split system.

Link : https://www.dash.org/forum/threads/...gh-performance-nodes.53374/page-7#post-232371

I guess Platform on all nodes is the only option to keep collateral at 1K, have a clear picture how things balance out, have IS working optimal and does not lead to centralization.

Or maybe the proposed 'Distributed Platform Storage' solution that just popped up does as well (as it avoids High Performance nodes and keeps 1K collateral unchanged), maybe worth taking a look at : https://www.dash.org/forum/threads/...ids-high-performance-nodes.53400/#post-232425
This solution certainly provide an alternative perspective / idea / concept on Platform storage.
 
Last edited:
Yeah, i think the 1K split system option would perhaps involve too much change to get it working. It would most likely need more masternodes to keep InstantSend working optimal over both assets (IS optimal for 1K split system means 2x 1920 nodes, while we currently only have 3701 active masternodes on the network, so too low to achieve that), which possibly means changing the 1K collateral to the down side to stimulate the setup of more masternodes. This is a change with a number of effects : it would affect VPS hosting fees, time interval between MN payments, number of MN payments for masternode owners. And the rewards from Platform and Core between both assets would also need to be structured in such a way, that it balance rewards out evenly over 1K Platform nodes and 1K normal nodes.

I am still confused how DCG research team came up with those ROI percentages for normal nodes and Platform nodes on the 1K split system. As that seems to have worked out how to balance the rewards part of the 1K split system.

Link : https://www.dash.org/forum/threads/...gh-performance-nodes.53374/page-7#post-232371

I guess Platform on all nodes is the only option to keep collateral at 1K, have a clear picture how things balance out, have IS working optimal and does not lead to centralization.

Or maybe the proposed 'Distributed Platform Storage' solution that just popped up does as well (as it avoids High Performance nodes and keeps 1K collateral unchanged), maybe worth taking a look at : https://www.dash.org/forum/threads/...ids-high-performance-nodes.53400/#post-232425
This solution certainly provide an alternative perspective / idea / concept on Platform storage.
Yeah q I’m just not buying it. The recommendation from dcg is we need fewer nodes to run platform. Lots of benefits, got it, that are slightly better than all running platform. Seems like an odd time to realize this, but it’s the world we live in. Now we’re told the only way to minimize nodes running platform is by increasing collateral and that there are no other ways to minimize nodes running platform.
 
QuantumExplorer
have you thought through a solution for dash platform that runs exclusively on transaction/platform fees instead of block reward? It seems like this provides good incentive, gives platform room to grow, and allows the market/network to prioritize it relative to demand.
 
Yeah q I’m just not buying it. The recommendation from dcg is we need fewer nodes to run platform. Lots of benefits, got it, that are slightly better than all running platform. Seems like an odd time to realize this, but it’s the world we live in. Now we’re told the only way to minimize nodes running platform is by increasing collateral and that there are no other ways to minimize nodes running platform.
running less nodes shouldn't be a problem. Provide no block reward, or very little block reward and you'll get less nodes. What am I missing?
 
can't platform node quorum selection, instead of being random, be related to the performance of the node? This way we incentivize high performance and don't need to worry as much about social engineering and playing with things as much
 
running less nodes shouldn't be a problem. Provide no block reward, or very little block reward and you'll get less nodes. What am I missing?
Honestly, I haven’t heard why this is such a problem. I tried watching the ama but just couldn’t get through it. Some will upgrade, some won’t. That’s how it always will be.

As a mno, I fully expect to have to upgrade my nodes on my own dime to run platform. I’m not bothered by this because I believe in the end state. I suspect others would do the same. You take that option away from me by messing with collateral.
 
Last edited:
@seanjae There are 2 1k options: Forcing everyone to run platform, or everyone has an option to run platform.

I was pointing out that the everyone has an option to run platform doesn't really work because there can not be an equilibrium between node types (hosting core + platform or just core).

UNLESS platform nodes no longer get core rewards, but that should lead to them no longer being required to support core.
 
QuantumExplorer
have you thought through a solution for dash platform that runs exclusively on transaction/platform fees instead of block reward? It seems like this provides good incentive, gives platform room to grow, and allows the market/network to prioritize it relative to demand.

Yes it was considered, but it seems somewhat of a hard method to bootstrap securely while maintaining decentralization.

Let's say the only rewards you get are from the fees of Platform. Initially these fees will be very very little while the hardware requirements will go way up. Even if you say you are going to run platform you might not pay for the hardware costs because it doesn't really make sense. Many people might not pay close attention to their node, because well it doesn't matter from an economic side to do so. This would lead most probably to quite a lot of issues.

Then we get into the problem that since we don't know how many nodes their will be (there is no optimal market equilibrium) we can't propose initial fee storage costs.

Another downside is that there is a high risk of centralization. As 1 big whale might run platform, when the total number of nodes doing so remains quite low.
 
Thanks @QuantumExplorer. I appreciate your time and effort to get platform across the line.

The part that’s tough to understand is the messaging around folks flocking to 4K/10k solutions. It seems to be unrealistic to me because it’s too cost prohibitive. But, if we go down the hpmn route, I hope you’re correct
 
@seanjae There are 2 1k options: Forcing everyone to run platform, or everyone has an option to run platform.

@QuantumExplorer
For the case of 1k/4k/10k platform nodes, you have made a lot of assumptions, and based on these assumption you presented (and choosed) a lot of options.

For example, we have two main/root roads.
1) Forcing every 1k node to run platform
2) Every 1k node has the option to run the platform or not.

Are the aforementioned two main/root options considered as your root?
If not, in what root/main option you rely on, in order to built your options and sub-options decision tree?
Among the two mutual excluding aforementioned main roads, which one do you choose?
What options and sub-options appear when you choose your prefered road?
And then, what additional sub-options appear and which among them are your prefered ones?

There is an (intentionaly hidden?) decision tree for our specific case, but for some strange mental reason you refuse to formalize it.
Could you please elaborate and present in a formal manner your assumptions and options?
 
Last edited:
Knipsel.JPG


Let's say the only rewards you get are from the fees of Platform. Initially these fees will be very very little while the hardware requirements will go way up. Even if you say you are going to run platform you might not pay for the hardware costs because it doesn't really make sense. Many people might not pay close attention to their node, because well it doesn't matter from an economic side to do so. This would lead most probably to quite a lot of issues.

Is the same not also the case with 10K HPM system ? The 10K HPM system gets 90% of its rewards from Platform and only 10% of its rewards from Core (from the blockrewards), so how does that matter any differently (from an economic point of view), with a system where 100% of its rewards comes from Platform ?
Does that 10% of the blockrewards from Core make that much difference for the 10K HPM system (from an economic point of view) ?

Source of that 90% rewards from Platform & 10% of the rewards from Core for the 10K HPM system : www.youtube.com/watch?v=zRj2cLjVR8Q&t=3s
(timestamp 39:35)

Also i would still like to know if the 1K split system ROI percentage as earlier calculated by DCG is still valid or not, or if it was miscalculated by your research team. Because those numbers clearly shows an equilibrium between node types (hosting core + platform or just core).

1K split system:
Masternodes: 6.4%
HPMasternodes: 6.6%

Source : https://www.dash.org/forum/threads/...gh-performance-nodes.53374/page-7#post-232371

Which means that what you have been saying about the 1K split system so far (no equilibrium between node types possible), directly clashes with what your own research team earlier calculated for the 1k split system. If that is the case, it would be very misleading from you to say the least.

I can understand InstantSend perhaps not running optimal on the 1K split system, but i do not understand why equilibrium between node types can not be reached, when your own research team already calculated an equilibrium between both assets.

I think currently the largest problem with Dash Platform is that it does not have a robust PoSe solution in place, where it knows which Platform nodes are properly participating or not, so the system knows which Platforms nodes to exclude from quorums. This is a very important part that is currently missing from Dash Platform and it seems to steer Dash Platform devs to centralized solutions and it is at the same time preventing a sharding solution.

Due to pressure to release this year, this very important part is getting skipped by Dash Platform devs (it would apparently need either 4 months or 6-12 months of additional coding).
See : https://www.dash.org/forum/threads/...ids-high-performance-nodes.53400/#post-232436

To me this all comes across as very poor planning over the years with regards to Dash Platform and its essentials to include at release.
Surely a robust PoSe scoring system for Dash Platform nodes should have been a priority from the start ?

Instead we have all these optimizations / recoding / reformatting / coding from scratch work done by devs over the years, while still lacking a robust PoSe scoring system for Dash Platform.
 
Last edited:
so how does that matter any differently (from an economic point of view), with a system where 100% of its rewards comes from Platform ?

You are comparing two very different things. In the first 50% of block rewards go to platform and HPMNs get 90% of their rewards from platform. The second 0% of the block rewards go to platform and HPMNs get 100% of their rewards from platform.

Let's take numbers:
Scenario 1 (10k): 2000 Dash produced per day with block rewards (just an example not accurate), 5 Dash generated in platform fees. 900 goes to platform. HPMNs collectively get 905 Dash from platform and 100 Dash from Core. Core nodes get 1000 Dash from core. Equilibrium is reached I believe around 182 HPMNs.

Scenario 2 (1k optional no split): 2000 Dash produced per day with block rewards (just an example not accurate), 5 Dash generated in platform fees. 0 goes to platform. Nodes running platform get 5 Dash from fees (100% of fees) normal Core rewards. There is no obvious equilibrium reachable because it is entirely fee based. Fees at launch will not be able to pay for hosting requirements.

Also i would still like to know if the 1K split system ROI percentage as earlier calculated by DCG is still valid or not, or if it was miscalculated by your research team. Because those numbers clearly shows an equilibrium between node types (hosting core + platform or just core).

This was a mistake in communication. In this scenario platform nodes are no longer being rewarded by core and therefore are not expected to perform services like chain locks/ IS. It is the only way we could get numbers for this scenario.

The issue here is that there are two ways to understand splits and they often can get confused. One way of understanding the split is that of how we are splitting the block reward. The second way of understand the split is what each type of node will make after the split. When we say 50/50 we mean the second type of split.
Nodes running platform + core will get 50% of rewards, and Nodes running just core will get 50% of rewards.

For the 10k split system this will mean 45% to platform and 55% to core. Each node type then makes 50%. We get to 45 with 50 - 1/10 * 50
For the 4k split system this will mean 37.5% to platform and 62.5% to core. Each node type then makes 50%. We get to 37.5 with 50 - 1/4 * 50
For the 1k split system this would mean 0% to platform and 100% to core. Each node type then makes 50%. We get to 0 with 50 - 1/1 * 50

When you reward platform at 0% things don't work.

If you were to instead reward platform at anything except 0% then you would get an equilibrium only if everyone were to run platform.
 
Last edited:
You are comparing two very different things. In the first 50% of block rewards go to platform and HPMNs get 90% of their rewards from platform. The second 0% of the block rewards go to platform and HPMNs get 100% of their rewards from platform.

Let's take numbers:
Scenario 1 (10k): 2000 Dash produced per day with block rewards (just an example not accurate), 5 Dash generated in platform fees. 900 goes to platform. HPMNs collectively get 905 Dash from platform and 100 Dash from Core. Core nodes get 1000 Dash from core. Equilibrium is reached I believe around 182 HPMNs.

Scenario 2 (1k optional no split): 2000 Dash produced per day with block rewards (just an example not accurate), 5 Dash generated in platform fees. 0 goes to platform. Nodes running platform get 5 Dash from fees (100% of fees) normal Core rewards. There is no obvious equilibrium reachable because it is entirely fee based. Fees at launch will not be able to pay for hosting requirements.

Scenario 2 (1k optional


This was a mistake in communication. In this scenario platform nodes are no longer being rewarded by core and therefore are not expected to perform services like chain locks/ IS. It is the only way we could get numbers for this scenario.

The issue here is that there are two ways to splits and they often can get confused. One way of understanding the split is that of how we are splitting the block reward. The second way of understand the split is what each type of node will make after the split. When we say 50/50 we mean the second type of split.
Nodes running platform + core will get 50% of rewards, and Nodes running just core will get 50% of rewards.

For the 10k split system this will mean 45% to platform and 55% to core. Each node type then makes 50%. We get to 45 with 50 - 1/10 * 50
For the 4k split system this will mean 37.5% to platform and 62.5% to core. Each node type then makes 50%. We get to 37.5 with 50 - 1/4 * 50
For the 1k split system this would mean 0% to platform and 100% to core. Each node type then makes 50%. We get to 0 with 50 - 1/1 * 50

When you reward platform at 0% things don't work.

If you were to instead reward platform at anything except 0% then you would get an equilibrium only if everyone were to run platform.

Thank you for the numbers / percentages and for the clarification with regards to the 1K split system ROI percentage, as earlier calculated by DCG
Did you mean to add something more to the 'Scenario 2 (1k optional' part that i highlighted in red ? Or is it there by mistake ?
At the moment it looks unfinished / cut off .....
 
Last edited:
Thank you for the numbers / percentages and for the clarification with regards to the 1K split system ROI percentage, as earlier calculated by DCG
Did you mean to add something more to the 'Scenario 2 (1k optional' part that i highlighted in red ? Or is it there by mistake ?
At the moment it looks unfinished / cut off .....
Oops just a mistake.
 
Oops just a mistake.
No problem.

Different topic :

I asked this on Dash Central to you just now and i also mentioned this as being a major problem for Dash Platform in my view, in one of my previous comments :

Why has no robust PoSe scoring solution for Dash Platform been developed so far ? Is it not an essential part of Dash Platform and was it also not planned for the initial release of Dash Platform ? If we had a robust PoSe scoring solution for Platform nodes in place (by having prioritized it in the past), then i assume a lot of the problems with large Dash Platform whales nodes stopping all at once during Dash Core software upgrades, causing quorums to stop, causing the Platform network to stop, would be mitigated ? Or more easily fixed ?

And there is the promotion of the 10K HPM system decision proposal by Sam, that makes it more easy to fix the above, when it is the devs themselves that created this problem for Dash Platform in the first place, by not prioritizing a robust PoSe scoring solution for Dash Platform in the past and by apparantly not willing to allocate more time for this, due to a pressure to release this year. Severly handicapping Dash Platform launch options and introducing more risk to Dash Platform upon release.

Some Platform teams are very close to finishing their tasks for Dash Platform and could therefore fully focus on developing a robust PoSe scoring solution for Platform... if developing a robust PoSe scoring solution for Dash Platform gets the priority it should have received years ago. Developing a robust PoSe scoring solution for Platform could even be subcontracted to Dash Incubator.

At this point i am starting to wonder if DCG has its priorities straight with regards to Dash Platform, and its upcoming release.

Should DCG release Dash Platform with all its essentials as communicated by DCG before, or should DCG skip an essential part and release Dash Platform due to pressure to release this year ? I would rather see Dash Platform release with all essentials (including a robust PoSe scoring solution for Dash Platform).

Also i am wondering about the limitations of Dash Platform on release, i think i read somewehere from Sam that Drive will not be used to store video's or pics.
Can we still use profile pics on Dash Platform, stored on Drive ? Or will that only be stored on our local computer / mobile phone / cloud space ?
 
Last edited:
Why has no robust PoSe scoring solution for Dash Platform been developed so far ? Is it not an essential part of Dash Platform and was it also not planned for the initial release of Dash Platform ? If we had a robust PoSe scoring solution for Platform nodes in place (by having prioritized it in the past), then i assume a lot of the problems with large Dash Platform whales nodes stopping all at once during Dash Core software upgrades, causing quorums to stop, causing the Platform network to stop, would be mitigated ? Or more easily fixed ?

A robust PoSe system should be designed to keep tabs on all Masternodes running platform. This needs to be done with game theory. As far as I know no project in the industry has yet built something like this. It is really complex to build systems like this. DCG can only do so much with the level of funding we have. We just do not enough people to do more than projects with x10 the funding. We can not be told to release asap but also to release the perfect system. Bitcoin was not perfect when it first came to market, and is not perfect now either, but it grew, just like Platform will grow.

And there is the promotion of the 10K HPM system decision proposal by Sam, that makes it more easy to fix the above, when it is the devs themselves that created this problem for Dash Platform in the first place, by not prioritizing a robust PoSe scoring solution for Dash Platform in the past and by apparantly not willing to allocate more time for this, due to a pressure to release this year. Severly handicapping Dash Platform launch options and introducing more risk to Dash Platform upon release.

I am ambivalent between 4K and 10K. I just want any solution that works in simulations and is safe. What would I or DCG stand to gain with 4K or 10K other that a system that best satisfies the needs of the network? I just want a good start to get people interested in the system that we have built. Also saying that the devs created this problem is also wrong. Other chains solve this problem with block rewards and that's what I had proposed for now.

There is nothing we could have cut to get PoSe out the door instead, and do you think the project can survive well if Platform is not out in the next 6 months?

Also i am wondering about the limitations of Dash Platform on release, i think i read somewehere from Sam that Drive will not be used to store video's or pics.
Can we still use profile pics on Dash Platform, stored on Drive ? Or will that only be stored on our local computer / mobile phone / cloud space ?

The plan here for Dashpay avatars was always to use a semi-centralized system here for release. What users do is that you set your avatar pic to the url of some service, then the multiple image processing servers take that picture and are responsible for its caching and distribution in various formats.
 
Back
Top