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

QuantumExplorer

Active Member
Dash Core Group
Aug 20, 2014
270
382
123
centralization of voting, the hijacking of the treasury
Can you elaborate how this is happening? Why does it matter if someone who has 4K Dash votes with 1 vote worth 4 = 1x4 points, or 4 votes worth 4 = 4x1 point?
I also don't see how this would be hijacking the treasury.

If Proof of Service is robust and a node
Proof of Service is not robust right now. I went over the fact that getting a robust proof of service for platform is a monumental task. I think we have a 30+ page document on how to build it. But as CTO I decided against building it for our MVP, basically because we can't delay Platform release another year.

Seven years in and a major design decision is thrust upon us imminently.
I have been CTO for about 1 year now. The decision to go with Tendermint based consensus was not one I made, actually I had a different approach but in the end I do like though where we are now and there is no way to sure my other solution would have been better or even would have worked, it was definitely more ambitious.

Back in the day of Evan, he seldom asked the network what they preferred. Andy and Bob never did. I do feel like there are two things both being said: one is that we should wait until we have more data, and the other is why did we wait so long? Which makes me think that this is really the right time. We are very close to feature freeze of the MVP and testing. Both GroveDB and Tenderdash teams should completely finish their MVP tasks this month, at which time they can be put to other teams to speed things up.

You are right about the fact that people might be unnerved that I'm proposing things that might change things up. But this was always going to happen as Platform is bringing about a lot of changes. Dash used to be the project of risk taking and bold calculated moves, then there was a slump, this last year however... has been pretty intense. And I feel like it's going to be a very exciting next year.
 
Last edited:
  • Like
Reactions: nik443 and xkcd

GrandMasterDash

Grizzled Member
Masternode Owner/Operator
Jul 12, 2015
3,423
1,459
1,183
Can you elaborate how this is happening? Why does it matter if someone who has 4K Dash votes with 1 vote worth 4 = 1x4 points, or 4 votes worth 4 = 4x1 point?
I also don't see how this would be hijacking the treasury.
Just to be sure, I think you are saying someone who owns four masternodes is no different to them owning one HPMN.

I have more than one node and I intentionally use different providers in different geographic and political jurisdictions. But I recognize others are perhaps more lazy and go with the same provider or location. But this is still not the problem.

The real problem is, you seem to hold on to a belief that collateral is preventing the option of a Sybil attack. That the accumulation of dash to attain 1K / 4K / 10K would send the price exponentially high. But for the last two years, the facts tell a different story. The overall masternode count has fallen while some node whales accumulated tens or hundreds of nodes, which comes with voting power (the inactivity of this power is of no consequence). And yet the dollar price for dash is currently less than $41. In the early years the collateral provided good defense against Sybil attacks, but I think the past two years has shows this is no longer the case.


So when you say the Proof of Service for Platform is not robust enough, that to me a major red flag. If we had a robust Proof of Service for both masternodes and HPMNs, then surely the collateral amount is of less importance. If the network has come to consensus that Node A is X and Y distance from Nodes B and C (fingerprinting latency and route etc), and nodes are retaining data and responding correctly to queries, then what else is there? Under such scenario, we could safely reduce basic collateral to, say, 100 dash. The same applies to any node type, including HPMNs, which would perhaps have a tighter criteria. I'm not saying it's easy to achieve but this is the crux of our current problem and more so if we allow HPMNs to slowly gentrify the network, where orange is the new black.
 
  • Like
Reactions: stan.distortion

vazaki3

Active Member
Jul 1, 2019
694
363
133
34
apogee.dynu.net
Dash Address
XnpT2YQaYpyh7F9twM6EtDMn1TCDCEEgNX
View attachment 11445

Me starting to feel like this picture, waiting on a reply from QuantumExplorer on the proposed 1K Masternode solution, that i reminded him two times already about. Maybe three times is a charm....

Link : https://www.dash.org/forum/threads/...gh-performance-nodes.53374/page-6#post-232345 (link to seanjae's orginal post is also in that link)
Why do you rely on QuantumExplorer's answer?
Who is QuantumExplorer and why his answer matters more than yours?
Why dont you add a forum poll asking the dashtalk members about your 1K Masternode solutuion?
Why dont you add a relevant governance question into the budget?



Me starting to feel like this picture, waiting for the Dash community members to decide by using polls, instead of begging individuals for decisions.
 
Last edited:

qwizzie

Grizzled Member
Aug 6, 2014
2,113
1,291
1,183
Why do you rely on QuantumExplorer's answer?
Who is QuantumExplorer and why his answer matters?
Why dont you add a forum poll for that?
Why dont you add a governance question?
Because i am asking CTO of DCG if this solution is technically workable, and if it indeed solves the security risk with Platform on all nodes.
Are you seriously asking me who Quantum Explorer is and why his answer matters?

Knipsel.JPG


Sam / Quantum Explorer is the one that is intiating this all. He is the one thats preparing all the DCG decision proposals with regards to this topic.
No point in doing a forum poll or add a governance question, untill i can get a response from a developer (or better yet from the CTO of DCG)
on the technical merits of this solution.
 
Last edited:

vazaki3

Active Member
Jul 1, 2019
694
363
133
34
apogee.dynu.net
Dash Address
XnpT2YQaYpyh7F9twM6EtDMn1TCDCEEgNX
Because i am asking CTO of DCG if this solution is technically workable, and if it indeed solves the security risk with Platform on all nodes.
Are you seriously asking me who Quantum Explorer is and why his answer matters?

View attachment 11447
Sam / Quantum Explorer is the one that is intiating this all. He is the one thats preparing all the DCG decision proposals with regards to this topic.
I know who QuantumExplorer is.
The fact that someone hired him as the Dash CTO (who did that?) says nothing to me, as long as nobody asked me whether I agree or not about it.
I am sure that among the DashCommunity members, there are individuals that fit better in CTO's role than QuantumExplorer does.
And I am more than sure that the collective wisdom of the DashCommunity members overcomes the wisdom of the best CTO we could ever had.

Thats why, add a poll , add a governance question.

No point in doing a forum poll or add a governance question, untill i can get a response from a developer (or better yet from the CTO of DCG)
on the technical merits of this solution.
Thats the root of your falacy. You dont have to wait for technical merits and disadvantages. You are asking a GOVERNANCE QUESTION.
If your question passes and a bounty is allocated for the implementation of it, let the developers and the designers to discover the technical solution, the merits and the disadvantages.
 
Last edited:

qwizzie

Grizzled Member
Aug 6, 2014
2,113
1,291
1,183
Thats the root of your falacy. You dont have to wait for technical merits and disadvantages. You are asking a GOVERNANCE QUESTION.
No, i am still trying to understand for myself if this 1K masternode solution is better then the Platform on all nodes solution or not.
Also with regards to DDoS attack vulnerability with these solutions (1K masternode / 4K HPM / 10K HPM), due to smaller masternode groups forming instead of having one large masternode group.
 
Last edited:

QuantumExplorer

Active Member
Dash Core Group
Aug 20, 2014
270
382
123
The reason why I haven't answered on the 1K solution proposed is that I don't just answer something quickly, it will take me hours to analyze this solution to be able to a good answer. When I say I don't have time... I think people are not realizing that all my days are just sleep/work/sleep/work. I had done something preliminary a while back and quickly discarded it. I'll have someone on our research team do this today, and I'll make sure you have your answers in the next 24 hours.
 
  • Like
Reactions: nik443 and qwizzie

QuantumExplorer

Active Member
Dash Core Group
Aug 20, 2014
270
382
123
The real problem is, you seem to hold on to a belief that collateral is preventing the option of a Sybil attack. That the accumulation of dash to attain 1K / 4K / 10K would send the price exponentially high. But for the last two years, the facts tell a different story. The overall masternode count has fallen while some node whales accumulated tens or hundreds of nodes, which comes with voting power (the inactivity of this power is of no consequence). And yet the dollar price for dash is currently less than $41. In the early years the collateral provided good defense against Sybil attacks, but I think the past two years has shows this is no longer the case.
I do not think I talked about the price... Analyses shows that we currently have the ability on our network for much more than the required 450 4K nodes. (450 is around where the equilibrium with the 1k nodes hits). To be clear I am not proposing this 4K solution because I believe it will raise the price of Dash directly. I am doing it because I believe it's a very good technical solution to a lot of the problems we are facing otherwise.

The sybil resistance of the system to the chain halt attack does not change all that much (maybe something like 64M$ to 42M$ - in a vacuum - would be much more expensive than this in open markets). But the upside of the system is that breaking Platform can never break stop Core features (IS/CL).

You did not explain how this would lead to a hijacking of the treasury?
 
  • Like
Reactions: nik443

QuantumExplorer

Active Member
Dash Core Group
Aug 20, 2014
270
382
123
Okay, @qwizzie I was able to quickly get some numbers for the 1K solution. It's 30 times less secure than the 4k Solution, while bringing down average rewards from 7% (4k Solution) to around 6.6% (because of the additional burden of higher hardware costs for the network on so many nodes - even though those nodes are 2 times less performant than what I put in the 4k Solution). Compared to all nodes running platform it's 2 times cheaper in fees for our users, but still very high compared to 4k solution. So all in all seems like there are no upsides.
 
  • Like
Reactions: nik443 and xkcd

qwizzie

Grizzled Member
Aug 6, 2014
2,113
1,291
1,183
Okay, @qwizzie I was able to quickly get some numbers for the 1K solution. It's 30 times less secure than the 4k Solution, while bringing down average rewards from 7% (4k Solution) to around 6.6% (because of the additional burden of higher hardware costs for the network on so many nodes - even though those nodes are 2 times less performant than what I put in the 4k Solution). Compared to all nodes running platform it's 2 times cheaper in fees for our users, but still very high compared to 4k solution. So all in all seems like there are no upsides.
1K Masternode solution 2x cheaper in fees then Platform on all nodes .. thats a pretty good deal if you ask me. That brings the Platform Storage fees to

7 cents for Base Identity
29 cents for setting up Identity + DPNS + DashPay
36 dollar for 1MB

Link : https://www.dash.org/forum/threads/...gh-performance-nodes.53374/page-3#post-232207

Correct ?

What about security risk ? (Platform bug taking out all masternodes) : does the 1K Masternode solution fix that security risk ?
What about DDoS vulnerability to this 1K masternode solution and also to 4K HPM / 10K HPM ?
 
Last edited:

qwizzie

Grizzled Member
Aug 6, 2014
2,113
1,291
1,183
@qwizzie but what are the upsides... it's 30 times less secure... In fact the security level would make this non viable.
In comparison to the 4K HPM solution (a centralized solution, due to higher collateral), yes it's 30 times less secure apparently

How secure is it in comparison to the Platform on all nodes solution ? (a decentralized solution with no change to collateral)
What about security risk ? (Platform bug taking out all masternodes) : does the 1K Masternode solution indeed fixes that security risk ?
What about DDoS vulnerability to this 1K masternode solution and also to 4K HPM / 10K HPM solutions ?

Why would this 1K masternode solution be non viable and the Platform on all nodes be viable, when comparing it directly to the 4K HPM solution ?
Which you are doing here... pretty much anything would be non viable that way.

Platform on all nodes solution and 4k HPM solution would be non viable in comparison to the 10K HPM solution then as well then, when taking this sort of reasoning with regards to security a step further.
 
Last edited:

QuantumExplorer

Active Member
Dash Core Group
Aug 20, 2014
270
382
123
The chance to be able to stop a quorum is 0.13% for the top MNO with 40% more Dash with the 1K solution. It's at 0.03% for the 4K solution and 0% with the 10k solution. So if you are interesting in decentralization for a security reason then you should prefer 10k over 4k over 1k. While this is counterintuitive it's still the truth. It's because of how probabilities work.

In my calculation, 1k running platform would need a 50$/month server. the 4k solution was using 100$/month servers. So yes the ddos protection and abilities of the 4k solution is stronger.

The everyone runs platform is secure as well, just because of the top node/total amount of nodes is small enough to not worry about this problem.
 
  • Like
Reactions: nik443 and xkcd

qwizzie

Grizzled Member
Aug 6, 2014
2,113
1,291
1,183
The chance to be able to stop a quorum is 0.13% for the top MNO with 40% more Dash with the 1K solution. It's at 0.03% for the 4K solution and 0% with the 10k solution.
So what is the percentage for Platform on all nodes solution ? I am seeing percentage for 1K solution (0.13%), i am seeing percentage for 4K solution (0.03%) and i am seeing percentage for 10K solution (0%). I am missing the percentage for Platform on all nodes solution. Also what are the cost for running Platform on all nodes per month on a server?

That way we can compare 1K masternode solution directly with the Platform on all nodes solution (Fees - Security - Server costs).
Fees are already 2x lower, so i am only wondering about security percentage (stopping a quorum) and server costs between these two solutions.
 
Last edited:

QuantumExplorer

Active Member
Dash Core Group
Aug 20, 2014
270
382
123
It's 0.0006% for this specific issue. However security will also be impacted by the strength of nodes. The 4k/10k solution will economically push for stronger nodes which will reduce the chance of ddos working.
 
  • Like
Reactions: nik443 and xkcd

qwizzie

Grizzled Member
Aug 6, 2014
2,113
1,291
1,183
Thank you for these calculations.

So to summarize :

Platform on all nodes solution (no change in collateral)

Storage Fees

14 cents for Base Identity
58 cents for setting up Identity + DPNS + DashPay
$72 for 1MB

Security

0.0006% chance to be able to stop a quorum

Server costs per month

$50

1K masternode solution (no change in collateral, but with a choice to run either a Platform node or a normal Classical node)

Storage Fees

7 cents for Base Identity
29 cents for setting up Identity + DPNS + DashPay
$36 for 1MB

Security

0.13% chance to be able to stop a quorum

Server costs per month

$50

4K HPM solution (increase in collateral)

Storage Fees

1.5 cents for Base Identity
6 cents for setting up Identity + DPNS + DashPay
$7.80 for 1MB

Security

0.03% chance to be able to stop a quorum

Server costs per month

$100

10K HPM solution (increase in collateral)

Storage Fees

0.5 cents for Base Identity
2 cents for setting up Identity + DPNS + DashPay
$2.60 for 1MB

Security

0% chance to be able to stop a quorum

Server costs per month

$150

Source : https://www.dash.org/forum/threads/...gh-performance-nodes.53374/page-3#post-232207 + above posts
 
Last edited:

qwizzie

Grizzled Member
Aug 6, 2014
2,113
1,291
1,183
The 1K masternode solution could be a pretty good compromise, for both DCG and the endusers / masternode owners.

Very reasonable storage fees
The higher server costs per month for the Platform 1K nodes will be compensated through Platform Credits after each epoch (18 days)
Offers masternode owners a choice between running a Classical node or a Platform node
Does not sacrifice too much on security (0.13% chance to be able to stop a quorum seems pretty acceptable)
Does not sacrice on decentralisation
Avoids a lot of unrest
 

QuantumExplorer

Active Member
Dash Core Group
Aug 20, 2014
270
382
123
For all nodes, you should put 50$/node too... I believe this will be the slightly higher than minimum level where we would start to get at least not a terrible level of service.

For the 10k system it's probably about 150$.

Okay something else then, let's imagine the current hardware requirements causes everyone to spend 20$ on their Masternodes each.

This means that the current yearly level of return is 6.8%

We can also give estimated rewards for the network in all these scenarios:



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%
1K split system: 6.49%
4K split system: 7.05%
10K split system 7.16%

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%
1K split system:
Masternodes: 6.4%
HPMasternodes: 6.6%
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%
1K split system:
Masternodes: 6.4%
HPMasternodes: 6.6%
4K split system:
Masternodes: 6.75%
HPMasternodes: 7%
10K split system:
Masternodes: 6.9%
HPMasternodes: 7.2%
 
Last edited:

QuantumExplorer

Active Member
Dash Core Group
Aug 20, 2014
270
382
123
@qwizzie I do not understand why a 1k solution would be better. It has the worst security of any system, the second highest fees, so very bad returns.

Why do believe this is a compromise scenario? Normal masternode owners would be much worse off, plus our system would run with half the performance.
 
  • Like
Reactions: nik443 and xkcd

qwizzie

Grizzled Member
Aug 6, 2014
2,113
1,291
1,183
We also have an ongoing miner-masternode reward-allocation shift going on untill 2025, could impact yearly return (or not)
See : https://www.dash.org/forum/threads/updated-mn-miner-reallocation-schedule.51453/

With regards to worst security of any system : they are all under 0.2% so what are we even talking about here with these very very low percentages ?

Storage Fees are cut in half in comparison with running Platform on all nodes (which was the most likely option for masternode owners against sacrificing decentralisation for lower fees, before the 1k masternode solution emerged). That is the compromise part for devs. Much lower storage fees for DCG worst case scenerio (with network not approving the DCG HPM decision proposals).

I can accept these storage fees and calculated ROI, when that means Dash stays decentralized and every masternode owner has the option to support Platform (and receive Platform rewards).

We have no way of knowing how Dash Platform adoptation will work out after launch. If we ever max out on our performance, due to overwhelming Platform adoptation then perhaps we can think of upgrading the hardware requirements a level higher ? Or start working on sharding ?

I also don't rule out fine-tuning and optimizing Dash Platform could bring those Storage fees a fair bit down over the years after launch.
Look at what you achieved already through optimization and looking at code again ? I think those storage fees were orginally a lot higher.
I bet a lot more eyes will be looking at this after launch.
 
Last edited:

QuantumExplorer

Active Member
Dash Core Group
Aug 20, 2014
270
382
123
0.13% per quorum, 24 quorums in a day, 356 days in a year (8544 tries).

1k Solution: P = (1 - (1 - 0.0013)^8544) = 99.99%

4k Solution: P = (1 - (1 - 0.0003)^8544) = 92.30%

10k Solution: P = (1 - (1 - 0.0000001)^8544) = 0.08% (best)

Now this is assuming you take the largest whale and add 40% to what they own.

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)

What we would most likely do though is rotate less often to increase the security of the 4k system, in the 1 k system there are so many nodes, that not rotating frequently could mean that some nodes get no platform rewards in an epoch as it is probabilistic to decide who is in a quorum.
 
  • Like
Reactions: nik443 and xkcd

qwizzie

Grizzled Member
Aug 6, 2014
2,113
1,291
1,183
0.13% per quorum, 24 quorums in a day, 356 days in a year (8544 tries).

1k Solution: P = (1 - (1 - 0.0013)^8544) = 99.99%

4k Solution: P = (1 - (1 - 0.0003)^8544) = 92.30%

10k Solution: P = (1 - (1 - 0.0000001)^8544) = 0.08% (best)

Now this is assuming you take the largest whale and add 40% to what they own.

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)

What we would most likely do though is rotate less often to increase the security of the 4k system, in the 1 k system there are so many nodes, that not rotating frequently could mean that some nodes get no platform rewards in an epoch as it is probabilistic to decide who is in a quorum.
What is the above for Platform on all nodes ?
P = (1 - (1 - 0.0006)^8544) i assume ? I tried to calculate, not sure i got the numbers right ( particularly not sure for the calculation with 20% more then they currently have). I got to 99.41 % so far (even that could be totally wrong)

And what is the above calculations from our current masternode quorums, running through our current masternodes ?
I am curious to those percentages as well, just to see what the current chance to stop a quorum right now is with current number of quorums, 356 days in a year

The whole thing is of course a lot of assumptions with regards to Dash whales possession and them that easily adding 20% or 40% to what they own and having whales attack Dash 356 days in a year (which seems a bit weird to take into account).
 
Last edited:

QuantumExplorer

Active Member
Dash Core Group
Aug 20, 2014
270
382
123
(1 - (1 - 0.0006)^8544) =99.41%

The whole point of v18 was to make instant send extremely extremely secure against double signs. That quorum has a lot of security features (rotating only a portion) that wouldn't benefit platform. It's really comparing apples to oranges, but I think we did the calculation and we are very close to 0.

The chance that the top MNO could double sign a Chain Lock is relatively high if he is able to attack the network with a timing attack, which is very very hard to pull off. In the case that they do pull this off, chain locks would just stop working and we would revert to pure PoW (basically they would gain absolutely nothing as the chain would continue). I guess at that point they could attempt a 51% hash based attack, but it's still a lot of ifs. We are going to improve the security of chain locks in v20.
 

qwizzie

Grizzled Member
Aug 6, 2014
2,113
1,291
1,183
So the only security weakness for the Platform on all nodes solution, 1K masternode solution and the 4K HPM solution would be timing attacks from large Dash whales ? (very very difficult to pull off). And security against that is getting beefed up through future v20

Only 10K HPM solution would have protection against this / very low chance of this happening, from the very start
Can this not be integrated into v19 ?
 
Last edited:

qwizzie

Grizzled Member
Aug 6, 2014
2,113
1,291
1,183
No @qwizzie, timing attacks are just for chain locks.
Ok, so if below calculations are for Platform and calculates the ability to stop the network and assumes we take the largest whale and add 40% to what they own
and ultimately come up with these numbers :

Platform on all nodes : P = (1 - (1 - 0.0006)^8544) =99.41%

1k Solution: P = (1 - (1 - 0.0013)^8544) = 99.99%

4k Solution: P = (1 - (1 - 0.0003)^8544) = 92.30%

10k Solution: P = (1 - (1 - 0.0000001)^8544) = 0.08% (best)

then that means something went really terribly wrong during Platform development over the many many years with regards to Platform security and we are now at a point, where only severe 10K centralization is the way forward. All other solutions (even the 4K HPM solution) are all extremely bad.

That is hard to believe. That would mean devs have really screwed up and should all be fired on the spot (my over the top reaction on an over the top situation)

Rotating less to increase security for 4K HPM's, while not being able to do the same for 1K masternodes or for Platform on all nodes (due to Platform Credits not being distributed deterministicly), is on itself a pretty bad design flaw of the probabilistic nature of the Platform Credits. In a way Dash Platform is in that regard one step forward and two steps back.

Anyways i think all these assumptions about Dash whales, attacking Platform 356 days in a year (8544 tries) with 40% more then what they own
is not all that realistic. It is nice to use as example to highlight a specific solution, but out in the real world this seems highly unlikely.


Maybe we should make percentages on above percentage numbers, to determine the odds of those percentage numbers coming true.
 
Last edited:
  • Like
Reactions: stan.distortion

Adding1kDash

New Member
Oct 9, 2022
13
22
3
36
All I am reading are just attempts to win the argument from QuantumExplorer, which I fully understand has performance/hardware geek myself. Which is okay if Dash was just a hobby project, but its not. It actually blocks the thought process of how to get as decentralized as possible, and move to how to make Dash the fastest. It is perhaps just a blindspot, or at worst childest unprofesional behavior, just like me overclocking, undervolting, killing all none required software, just to get that extra low Watt usage or higher fps, during benchmarking.

For example the argument that masternodes are running less than optimal server is not one and the same question.
that is actually about more and stricter PoSe requirements, the options to test if the hardware and network connections are optiomal actually increase with dashplatform.

"just an example"
You could for example give all Dashplatform nodes the option to querry idle other masternodes, if they don't act fast enough, the get there first PoSe points, than just hammering them till they get a PoSe ban, the optimal masternodes will not suffer because the can handle the let say 50 querry request per second. Obviously, the test would require, 3 party's one the dashplatform node, 2 the challanger(the one claiming that a specific node is running below required specs), and 3 observer that both party's are playing fair.