• 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 ?

Status
Not open for further replies.
You understood correctly. QE and i looked at scenarios that are close to extreme. So yes, if a whale owned 100 nodes, then we'd have calculated for 120/140 nodes.
 
You understood correctly. QE and i looked at scenarios that are close to extreme. So yes, if a whale owned 100 nodes, then we'd have calculated for 120/140 nodes.

Looking at QE's answer the whale owning 20% more (instead of 40% more) then he currently has, leads to the following percentages :

Let's do the same calculation with 20% more than they currently have

1k Solution: P = (1 - (1 - 0.0013)^8544) = 38.66%

4k Solution: P = (1 - (1 - 0.0003)^8544) = 5.84%

10k Solution: 0 (best, +20% can never give them ability to stop network)

source : https://www.dash.org/forum/threads/...gh-performance-nodes.53374/page-7#post-232375

Is that correct ? Or do we only use the percentages as mentioned here ? --> https://www.dash.org/forum/threads/...h-performance-nodes.53374/page-10#post-232464 (and calculate them further over the entire time frame)

Maybe it is better to get some verification of the percentage numbers on both 20% more nodes for whales and 40% more nodes for whales, calculated over the entire time frames for each solution. For easy comparising. I feel like i am comparing apples to oranges right now.

Current proposed solutions : Platform on all nodes, 4K HPM system, 10K HPM system.
 
Last edited:
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.
There is a technical limit to what is optimal for performance of the currently implemented Platform (Evo) while using acceptable number of nodes for decentralization. Mentioned was something around 200. So take 200 as the target number of nodes and propose the initial fee storage costs from that 200 count. Cap the number of possible Evo nodes to 200 - there cannot be more.

Keep Evo nodes out from getting core rewards, which probably means they no longer need to support core. In other words, anybody can start running Evo node and compete for being in the group of 200. That is the most open scenario for competition. In my opinion that is OK to start with - to see the demand of the network to have Evo up and running. This option does not require any voting, because it will be non-coercive upgrade as usual associated with a software package for Evo node. The vote is running an Evo node.

For sake of security though it might be required to prove ownership of 1, 2, 3 or more master-nodes in order to run Evo node. That may or may not be introduced later. The requirement of proving MN ownership would provide some level of decentralization (or at least its analysis) based on the MN network.

For the performance sake there must be a way to measure / evaluate the performance - there is no way around that, no matter the size or juiciness of a "carrot". Start with some trivial evaluation, such as reading /proc/cpuinfo and /proc/meminfo to test the machine before installation and at every start-up. Obviously DCG would later have to figure out some mechanism how an Evo node can be evaluated for performance from other machine over the net. And thus keep up the competition for nodes performing bellow certain threshold level by being potentially replaced with a new candidate node of a sufficient performance.
 
Last edited:
I will answer tomorrow

Also something else for tomorrow :

1: If Tendermint has a deterministic protocol and TenderDash is a fork of Tendermint, why are the Platform rewards not deterministic of nature as well ? Why are they probabilistic of nature ? (causing a certain degree of uncertainty in their payouts after an epoch)

2 : If we look at your statement from one of your previous posts :

Knipsel.JPG


How come estimates for rewards gives HP masternodes on the 4K HPM and 10K HPM solution a higher annual yield return, then normal masternodes ?

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

At network start (no fees generated) AND Assuming platform nodes get slightly more rewards (educated guesses on how the system will stabilise).
All nodes run platform: 6.11%

4K split system:
Masternodes: 6.9%
HPMasternodes: 7.2%

10K split system:
Masternodes: 7%
HPMasternodes: 7.3%

At network start (no fees generated) AND Assuming platform nodes get slightly more rewards AND platform nodes run stronger hardware to better server the network and ensure their profits(educated guesses on how the system will stabilize and hardware used).

All nodes run platform: 6.11%

4K split system:
Masternodes: 6.75%
HPMasternodes: 7%

10K split system:
Masternodes: 6.9%
HPMasternodes: 7.2%

These percentage numbers on estimated annual yield from DCG research team and your own statement that having a separate set of node does not put whales at an economic advantage, seem to mismatch. There does seem to be an economic advantage for whales (although a very small one).

3 : Decentralisation seems to have a different meaning for Dash users and for Dash developers.

For Dash users (personal assumption based on comments in this thread) decentralisation means keeping as high a number of nodes as possible, as acccessible as possible. Increasing collateral and creating a much smaller group of 'Super masternodes / High Performance nodes' will compromise this and leads to a situation where Dash can be very easily criticized by its direct oppponents and future Dash marketing efforts can be very easily undermined.
For Dash developers (Sam's statement mainly) decentralization is important, so the project can remain secure against bad or incompetent actors

I think it is important that both aspects of decentralisation should be taken into consideration by Dash developers, not just the last.
 
Last edited:
The subject is quite simple.
Increasing collateral means decreasing the number of entities involved in quorums. That's an act of centralization by pure definition. So the question is: do we want to trade some decentralization for better perfomance, lesser fees, higher speed and easier coding?
My answer is NO.
Take your choice, guys.
 
Looking at QE's answer the whale owning 20% more (instead of 40% more) then he currently has, leads to the following percentages :



source : https://www.dash.org/forum/threads/...gh-performance-nodes.53374/page-7#post-232375

Is that correct ? Or do we only use the percentages as mentioned here ? --> https://www.dash.org/forum/threads/...h-performance-nodes.53374/page-10#post-232464 (and calculate them further over the entire time frame)

Maybe it is better to get some verification of the percentage numbers on both 20% more nodes for whales and 40% more nodes for whales, calculated over the entire time frames for each solution. For easy comparising. I feel like i am comparing apples to oranges right now.

Current proposed solutions : Platform on all nodes, 4K HPM system, 10K HPM system.

off topic, but I did say it was extreme, that was poor wording, it's cautious. If the first whale is an exchange, then it's better to expect their number of nodes to slowly grow i believe.

I will get back to you in a bit on the security indeed because it feels that there are many numbers which have been thrown around and while they seem to be in the good ranges, it's hard for me to tell which parameters were used, so indeed it does not seem proper. Give me some time to answer to the others first then i'll get to looking at it.

1: If Tendermint has a deterministic protocol and TenderDash is a fork of Tendermint, why are the Platform rewards not deterministic of nature as well ? Why are they probabilistic of nature ? (causing a certain degree of uncertainty in their payouts after an epoch)

The answer will disappoint you: It's simple and reliable enough for a release. Without PoSe it's hard to make a proper and fair reward system and basing it off of some "verifiable" information is, right now, the best solution. We can also make a neat system but then add months of delay to Platform, i'm sure none of us want it.

2 : If we look at your statement from one of your previous posts ... How come estimates for rewards gives HP masternodes on the 4K HPM and 10K HPM solution a higher annual yield return, then normal masternodes ?

I am not sure what you are asking, there are three scenarios, two of which where the HPM get slightly more yield. Are you asking why we chose to model the stabilisation of the yield at such values? Because there are social factors at play on top of the economical ones. Upgrading takes some efforts, might be scary as it's a change, if your node goes down it's scarier... There is a certain value gain where probably most nodes if not all would simply not bother doing the upgrade for. As to what that value is if it exists, i do not know, Rion made the bulk of calculations and then QE looked at what happens if the yield have such a slight difference. Hence the "educated guess" part.
 
There is a technical limit to what is optimal for performance of the currently implemented Platform (Evo) while using acceptable number of nodes for decentralization. Mentioned was something around 200. So take 200 as the target number of nodes and propose the initial fee storage costs from that 200 count. Cap the number of possible Evo nodes to 200 - there cannot be more.

Keep Evo nodes out from getting core rewards, which probably means they no longer need to support core. In other words, anybody can start running Evo node and compete for being in the group of 200. That is the most open scenario for competition. In my opinion that is OK to start with - to see the demand of the network to have Evo up and running. This option does not require any voting, because it will be non-coercive upgrade as usual associated with a software package for Evo node. The vote is running an Evo node.

For sake of security though it might be required to prove ownership of 1, 2, 3 or more master-nodes in order to run Evo node. That may or may not be introduced later. The requirement of proving MN ownership would provide some level of decentralization (or at least its analysis) based on the MN network.

For the performance sake there must be a way to measure / evaluate the performance - there is no way around that, no matter the size or juiciness of a "carrot". Start with some trivial evaluation, such as reading /proc/cpuinfo and /proc/meminfo to test the machine before installation and at every start-up. Obviously DCG would later have to figure out some mechanism how an Evo node can be evaluated for performance from other machine over the net. And thus keep up the competition for nodes performing bellow certain threshold level by being potentially replaced with a new candidate node of a sufficient performance.

I will pass on your advice to QE, i think he will appreciate it as it is very ground to earth thoughts.
 
I am not sure what you are asking, there are three scenarios, two of which where the HPM get slightly more yield. Are you asking why we chose to model the stabilisation of the yield at such values? Because there are social factors at play on top of the economical ones. Upgrading takes some efforts, might be scary as it's a change, if your node goes down it's scarier... There is a certain value gain where probably most nodes if not all would simply not bother doing the upgrade for. As to what that value is if it exists, i do not know, Rion made the bulk of calculations and then QE looked at what happens if the yield have such a slight difference. Hence the "educated guess" part.

No, i am more interested in this part of my question to you (although it is not really framed as a question, more as an observation) :

These percentage numbers on estimated annual yield from DCG research team and your own statement that having a separate set of node does not put whales at an economic advantage, seem to mismatch. There does seem to be an economic advantage for whales (although a very small one).

This is by looking at the percentage numbers provided by QE (obtained i assume, from his research team).
See : https://www.dash.org/forum/threads/...gh-performance-nodes.53374/page-7#post-232371

I highlighted that economic advantage in green in my previous post.
 
Last edited:
The subject is quite simple.
Increasing collateral means decreasing the number of entities involved in quorums. That's an act of centralization by pure definition. So the question is: do we want to trade some decentralization for better perfomance, lesser fees, higher speed and easier coding?
My answer is NO.
Take your choice, guys.

that's why we're polling, it's a divisive topic. Thus it is good that you speak your mind.

My text below is also an answer to qwizzie.

I do, however, disagree. The main features of decentralization are not just on and off, they are gradual qualities. And as with everything in life it's not all black or white, decentralization also has downsides such as more data replication, having the edges of the network struggle to keep in line with the rest on consensus and so on. A right balance has to be striken.

As qwizzie stated, the community cares about having a lot of nodes. What i want to point out is the following:
You care about the number of nodes, not because of the number itself but because of what it implies. You wish decentralization and you will agree with me that a right balance has to be striken: if you have a billion nodes in your system for 10 or so users well... more decentralization will be counterproductive.

How do you then estimate where such a balance lies? You look at metrics that are directly decided by the amount of nodes. Metrics such as ensured censorship resistance. I personally believe that there will not be, on release of Platform, such issue with a smaller amount of beefier nodes and it is a post release issue. Also, as stated, if censorship do happen and some Platform nodes go down, a free market situation does enable other nodes in another country to simply take over the now increased yield.

Another two points which are, in my opinion, very important:
  1. If all the nodes support Platform and a state order falls in say... America to not support Platform, all nodes at AWS for example will go down. A lot of MNs are hosted at AWS right, qwizzie? This feels to me like a very unsafe possibility. And it has happened to other projects in the recent past afaik. ^ this is in my opinion perhaps the most important thing to consider with this poll. Forcing all nodes to participate does not make it more censorship resistant but instead makes the network weaker. ^
  2. This number of nodes can be tweaked in the future to what suits the network best.
 
  1. If all the nodes support Platform and a state order falls in say... America to not support Platform, all nodes at AWS for example will go down. A lot of MNs are hosted at AWS right, qwizzie? This feels to me like a very unsafe possibility. And it has happened to other projects in the recent past afaik. ^ this is in my opinion perhaps the most important thing to consider with this poll. Forcing all nodes to participate does not make it more censorship resistant but instead makes the network weaker. ^
  2. This number of nodes can be tweaked in the future to what suits the network best.

What if the 10K HPM solution is voted in and it turns out that most of the 10K HPM's are setup through AWS ?
Same problem, correct ? Maybe even more of a problem as this group is much smaller....

Anyways, this is the current state (pretty decently decentralized at the moment)

Knipsel.JPG


Source : https://chainz.cryptoid.info/dash/#!network
 
Last edited:
No, i am more interested in this part of my question to you (although it is not really framed as a question, more as an observation) :

These percentage numbers on estimated annual yield from DCG research team and your own statement that having a separate set of node does not put whales at an economic advantage, seem to mismatch. There does seem to be an economic advantage for whales (although a very small one).

This is by looking at the percentage numbers provided by QE (obtained i assume, from his research team).
See : https://www.dash.org/forum/threads/...gh-performance-nodes.53374/page-7#post-232371

I highlighted that economic advantage in green in my previous post.

Let me reformulate then, as i answered that point but i think we misunderstood each others. There is nothing inherent to the system or design that put whales at a benefit. The only reason that there might be a difference in yield is social factors. The system is fair but IF whales see a possibility to increase yield from 6.9% to 7.2% and do not take it because it is too much of a bother, then the yield stabilizes there. Will it happen? Maybe, Maybe not. Would social factors stabilize the yields at those rates? Maybe, Maybe not. That's why an educated guess is called like that, it's a guess.

It's like saying to a group of people: all of you who bend down can get a share of this bill of 10 dollars/euros on the ground. One person does it, cool they get 10. If 19 people already bend, do you bend to get 1/20th of the bill = 50 cents? and if 99 people bend, do you bend down for 10 cents?

Do you understand why it is both true that the system does not create a financiary advantage toward whales while simultenaously enabling the possibility that a minimal gap in yields might exists?

Once again, it is speculations to show the communities values to base their opinions off.
 
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.

No it is not, the trigger that made Satoshi nakamto build bitcoin was because of the abuse of the centralized powers during 2008 banking crisis.
It seems like Dash its leadership has lost it's roots completely, decentralisation just seems like a marketing term at this point, and Dash is really not going to get away with it with just 180HP masternodes, 450 does not sound much better either.

If you study some ancient history you would know that people where able to all work together in community's round about 100 people.
Everyone that has lived in small villages knows that everone knows every one. As these villages grow people start to lose track, I can tell from my own experciences and the people around me (I have lived in a village all my life, and have friends in multiple villages around me). That as that numbers goes up it's hard to keep track, once it's gets into a couple of 1000's this no longer is the case.

This is where Evan Duffield was aiming at when he decided on the number a 1000Dash. A number to high to cause centralisation, but low enough that it would still be fast enough to serve the whole world one day.

Stop thinking with just your logical part of the brain, use your emotional part of the brain, research and realize that what I am saying has been proven to be truth time and time again.
 
What if the 10K HPM solution is voted in and it turns out that most of the 10K HPM's are setup through AWS ?
Same problem, correct ? Maybe even more of a problem as this group is much smaller....

Anyways, this is the current state (pretty decent decentralized at the moment)

View attachment 11460

Source : https://chainz.cryptoid.info/dash/#!network


Thanks for the link, i appreciate it.

If 10K or 4K nodes are all geolocalized and get taken down by whatever, as long as there is data replication somewhere, not everything is lost and new nodes begin anew. I don't believe there being 400 or 500 nodes will change much to the geolocalisation in general if you ask me. I am ready to be the distribution will stay the same roughly. Do notice though that i don't have a strong opinion on 4K vs 10K or whatever K. If you ask me all that matters is that not all nodes provide the service so we can get down to better fees and a protection against worst case scenario as stated in my previous message. If Platform is banned, it must not impact the Core chain.
 
If Platform is banned, it must not impact the Core chain.

In a way it would be much more difficult to ban Dash Platform on the Platform on all nodes solution (you would need to ban Dash as cryptocurrency everywhere), then to ban Dash Platform on the 4K HPM solution or the 10K HPM solution (you can just single out HPM's).
 
Also, as stated, if censorship do happen and some Platform nodes go down, a free market situation does enable other nodes in another country to simply take over the now increased yield.

That's exactly what I fear most: censorship happen and nodes decided to censor the network ARE NOT going down.

But thank you for your answer. Yes, indeed, there is a certain point of balance in centralization / decentralization. Right now there are about 1000 of entities validating the network through PoS and unknown number of miners doing their job with PoW. With HPMNs Platform will only have a few or several dozens of PoS validators - that's the same with market right now and absolutely not enough, as it seems to me.
 
No it is not, the trigger that made Satoshi nakamto build bitcoin was because of the abuse of the centralized powers during 2008 banking crisis.

I am not talking about why it is made. I am talking of how. I will directly quote the original paper by Satoshi:
1666081987935.png


This, above, is textbook game theory. I will paste the first line of wikipedia to make sure we understand the same thing: "Game theory is the study of mathematical models of strategic interactions among rational agents. "

It seems like Dash its leadership has lost it's roots completely, decentralisation just seems like a marketing term at this point, and Dash is really not going to get away with it with just 180HP masternodes, 450 does not sound much better either.

We are polling the community to know what the community want. We haven't taken a decision to push 180 or 450 nodes. If anything, it proves that we are trying to do our best to stick to the community's values if you ask me.


This is where Evan Duffield was aiming at when he decided on the number a 1000Dash. A number to high to cause centralisation, but low enough that it would still be fast enough to serve the whole world one day.

This was the right balance in his opinion for this purpose, we now ask you what your right balance is for a Platform. I hope you appreciate that we put the community back in the center with this decision.

Stop thinking with just your logical part of the brain, use your emotional part of the brain, research and realize that what I am saying has been proven to be truth time and time again.

I am not going to shut off my brain thinking part, else you might get some poor security in the future :b
 
In a way it would be much more difficult to ban Dash Platform on the Platform on all nodes solution (you would need to ban Dash as cryptocurrency everywhere), then to ban Dash Platform on the 4K HPM solution or the 10K HPM solution (you can just single out HPM's).

That's up for debate, i'm not sure you should especially debate with me out of everyone, but my opinion is that the danger to Core is not worth the small (small according to me) increase in censorship resistance. I think it's almost exactly the same security in that regard if you get all nodes in or a subset. It's a game of possibles and of when we want to increase security depending on socio-economical factors, later on. We don't have an unlimited budget sadly, but otherwise those questions would not even be questions.

As long as one node which holds the chain is not in a censored country, the network bounces.
 
So has consensus been reached among DCG with regards to HPM's ? There was an internal poll going on that had some initial votes from devs, but because it was weekend that was just the initial outcome. What was the final outcome ?

The reason i ask is because Sam mentioned previously that a decission proposal would most likely be prepared by DCG, if there was internal DCG consensus for a HPM solution.

See : https://www.dash.org/forum/threads/...gh-performance-nodes.53374/page-3#post-232226
 
Last edited:
That's exactly what I fear most: censorship happen and nodes decided to censor the network ARE NOT going down.

If a miner proposes a block including something a major part of validators wish to censor, then they will do so, and having 400 or 100 will not change the needed ratio, nor will it imply nodes being taken down or not.

But thank you for your answer. Yes, indeed, there is a certain point of balance in centralization / decentralization. Right now there are about 1000 of entities validating the network through PoS and unknown number of miners doing their job with PoW. With HPMNs Platform will only have a few or several dozens of PoS validators - that's the same with market right now and absolutely not enough, as it seems to me.

A small correction, It'd would be estimated in the numbers of 200-400. Not a dozen
 
Status
Not open for further replies.
Back
Top