Author Topic: Masternodes Testing (Round 2)  (Read 137143 times)

georgem

  • Tech Admin
  • ******
  • Posts: 948
    • View Profile
Re: Masternodes Testing (Round 2)
« Reply #315 on: February 13, 2015, 06:21:45 pm »
Evan Duffield has been struggling with off-chain voting for 6+ months - it's still not reliable enough and needs a centralised node to ensure even payment distribution. I think Mr Spread's on-chain solution is better.

I like to phantasize about the blockchain being like a living machine, and that it has this on-chain short-time-memory that let's it remember the state of the current election/de-election process...

I think it's generally best practice for cryptocurrencies to involve the blockchain as much as possible, since it can be considered "the incorruptible brain" of the network.

georgem

  • Tech Admin
  • ******
  • Posts: 948
    • View Profile
Re: Masternodes Testing (Round 2)
« Reply #316 on: February 13, 2015, 06:48:44 pm »
We have 1823 in the list.
903 are elected.

It's funny that not only has no <150er mn ever been elected...
now we don't even see any <200er mn ever being able to get elected.

It's like this barrier is being increased slowly over time.

So I guess whoever reads this and has mn with lower than 400 SPR in them, please stop them, and merge them to 400er MNs...

This way we will get to our goal of 1440 mns much faster.

MyFarm

  • Sr. Member
  • ****
  • Posts: 259
    • View Profile
Re: Masternodes Testing (Round 2)
« Reply #317 on: February 13, 2015, 07:16:55 pm »
We have 1823 in the list.
903 are elected.

It's funny that not only has no <150er mn ever been elected...
now we don't even see any <200er mn ever being able to get elected.

It's like this barrier is being increased slowly over time.

So I guess whoever reads this and has mn with lower than 400 SPR in them, please stop them, and merge them to 400er MNs...

This way we will get to our goal of 1440 mns much faster.
There's no need to stop lower SPR MN's.  Adding higher SPR MN's one get us to our goal faster.  At present, all the higher SPR MN's will be elected first THEN it will move to the lower priced MN's and elect them until we have 1440.  THEN we start adding (and removing) higher SPR MN's to testing the competition system.

georgem

  • Tech Admin
  • ******
  • Posts: 948
    • View Profile
Re: Masternodes Testing (Round 2)
« Reply #318 on: February 13, 2015, 07:22:43 pm »
We have 1823 in the list.
903 are elected.

It's funny that not only has no <150er mn ever been elected...
now we don't even see any <200er mn ever being able to get elected.

It's like this barrier is being increased slowly over time.

So I guess whoever reads this and has mn with lower than 400 SPR in them, please stop them, and merge them to 400er MNs...

This way we will get to our goal of 1440 mns much faster.
There's no need to stop lower SPR MN's.  Adding higher SPR MN's one get us to our goal faster.  At present, all the higher SPR MN's will be elected first THEN it will move to the lower priced MN's and elect them until we have 1440.  THEN we start adding (and removing) higher SPR MN's to testing the competition system.

I fear that that's exactly what's not going to happen.

The bottle-neck of 10 votes per block is completely and constantly used up not only by the higher mns getting elected, but primarily by the higher mns constantly getting votes for their de-election.

I think the way the system is designed now causes small mns to be completely outmaneuvered.

Just look at the list, pretty much all higher mns ARE elected. How many of them are not elected? maybe 10%?
So how come all the <=200 have yet to get a single election? (with the exception of the small mns that were active in the very beginning of this test round and were able to get elected when the barrier was low).

MyFarm

  • Sr. Member
  • ****
  • Posts: 259
    • View Profile
Re: Masternodes Testing (Round 2)
« Reply #319 on: February 13, 2015, 07:36:27 pm »
Just look at the list, pretty much all higher mns ARE elected. How many of them are not elected? maybe 10%?
So how come all the <=200 have yet to get a single election? (with the exception of the small mns that were active in the very beginning of this test round and were able to get elected when the barrier was low).
They won't get an election until the higher SPR MN's are all elected.  Let's let the higher ones get elected and then allow the lower ones to get elected.  IF they don't, then that's a bug Mr. Spread needs to fix.  But I suspect once all the higher SPR MN's are elected, the system will move to the lower ones.

Strumpet!

  • Full Member
  • ***
  • Posts: 158
    • View Profile
Re: Masternodes Testing (Round 2)
« Reply #320 on: February 13, 2015, 07:40:22 pm »
We have 1823 in the list.
903 are elected.

It's funny that not only has no <150er mn ever been elected...
now we don't even see any <200er mn ever being able to get elected.

It's like this barrier is being increased slowly over time.

So I guess whoever reads this and has mn with lower than 400 SPR in them, please stop them, and merge them to 400er MNs...

This way we will get to our goal of 1440 mns much faster.
There's no need to stop lower SPR MN's.  Adding higher SPR MN's one get us to our goal faster.  At present, all the higher SPR MN's will be elected first THEN it will move to the lower priced MN's and elect them until we have 1440.  THEN we start adding (and removing) higher SPR MN's to testing the competition system.

I fear that that's exactly what's not going to happen.

The bottle-neck of 10 votes per block is completely and constantly used up not only by the higher mns getting elected, but primarily by the higher mns constantly getting votes for their de-election.

I think the way the system is designed now causes small mns to be completely outmaneuvered.

Just look at the list, pretty much all higher mns ARE elected. How many of them are not elected? maybe 10%?
So how come all the <=200 have yet to get a single election? (with the exception of the small mns that were active in the very beginning of this test round and were able to get elected when the barrier was low).

You are getting exactly what you wanted, although the process is slower than anticipated. I told you, there are never going to be any 100 SPR MNs on mainnet. There are very quickly going to be 1440 @ at least 1000 SPR each.

def15

  • Jr. Member
  • **
  • Posts: 62
    • View Profile
Re: Masternodes Testing (Round 2)
« Reply #321 on: February 13, 2015, 07:40:42 pm »
We have 1823 in the list.
903 are elected.

It's funny that not only has no <150er mn ever been elected...
now we don't even see any <200er mn ever being able to get elected.

It's like this barrier is being increased slowly over time.

So I guess whoever reads this and has mn with lower than 400 SPR in them, please stop them, and merge them to 400er MNs...

This way we will get to our goal of 1440 mns much faster.
There's no need to stop lower SPR MN's.  Adding higher SPR MN's one get us to our goal faster.  At present, all the higher SPR MN's will be elected first THEN it will move to the lower priced MN's and elect them until we have 1440.  THEN we start adding (and removing) higher SPR MN's to testing the competition system.

I fear that that's exactly what's not going to happen.

The bottle-neck of 10 votes per block is completely and constantly used up not only by the higher mns getting elected, but primarily by the higher mns constantly getting votes for their de-election.

I think the way the system is designed now causes small mns to be completely outmaneuvered.

Just look at the list, pretty much all higher mns ARE elected. How many of them are not elected? maybe 10%?
So how come all the <=200 have yet to get a single election? (with the exception of the small mns that were active in the very beginning of this test round and were able to get elected when the barrier was low).

I ended some 100spr MN's and turned them into 800spr MN's, just to get elected quick.

So as my 800spr MN's came back online they took priority over 10 other lower priced MN's.

100,000 tspr are mined everyday, and people are constantly adding high value spr MN's, cutting out the lower ones from election.

The supply on testnet is ridicules now... Must be around 500,000 now? And mostly to tspr testers.

MyFarm

  • Sr. Member
  • ****
  • Posts: 259
    • View Profile
Re: Masternodes Testing (Round 2)
« Reply #322 on: February 13, 2015, 07:46:12 pm »
You are getting exactly what you wanted, although the process is slower than anticipated. I told you, there are never going to be any 100 SPR MNs on mainnet. There are very quickly going to be 1440 @ at least 1000 SPR each.

What makes you think we'll have 1440 at 1k each?  That takes up much of the coin supply.

Strumpet!

  • Full Member
  • ***
  • Posts: 158
    • View Profile
Re: Masternodes Testing (Round 2)
« Reply #323 on: February 13, 2015, 07:50:59 pm »
You are getting exactly what you wanted, although the process is slower than anticipated. I told you, there are never going to be any 100 SPR MNs on mainnet. There are very quickly going to be 1440 @ at least 1000 SPR each.

What makes you think we'll have 1440 at 1k each?  That takes up much of the coin supply.

Current coin supply, yes. Three months or six months from now with ~260000 new SPR minted each month... not so much. Forget about MNs @ under 1000 SPR.

MyFarm

  • Sr. Member
  • ****
  • Posts: 259
    • View Profile
Re: Masternodes Testing (Round 2)
« Reply #324 on: February 13, 2015, 07:54:07 pm »
You are getting exactly what you wanted, although the process is slower than anticipated. I told you, there are never going to be any 100 SPR MNs on mainnet. There are very quickly going to be 1440 @ at least 1000 SPR each.

What makes you think we'll have 1440 at 1k each?  That takes up much of the coin supply.

Current coin supply, yes. Three months or six months from now with ~260000 new SPR minted each month... not so much. Forget about MNs @ under 1000 SPR.

Ah, ok.  Your, "quickly" and my, "quickly" were different durations :)

georgem

  • Tech Admin
  • ******
  • Posts: 948
    • View Profile
Re: Masternodes Testing (Round 2)
« Reply #325 on: February 13, 2015, 07:56:35 pm »
You are getting exactly what you wanted, although the process is slower than anticipated. I told you, there are never going to be any 100 SPR MNs on mainnet. There are very quickly going to be 1440 @ at least 1000 SPR each.

Bullfrog.

I didn't want high amount MNs to be preferred by the system when it comes to the ability of giving and taking votes.
I don't even know why Mr. Spread decided to use a formula that treats them better than the rest.

WTF?

Why isn't this completely random?

Let's look at how masternodes are elected:

How masternodes are elected?
Each node monitor the network and assign scores to each masternode. This scores depend on how well masternodes provide their services, for instant transactions this will be time delay between transaction and its confimation by masternode. Since there may be no transactions and only elected masternodes will confirm transactions there is an empty service - each masternode will broadcast messages signining certain blocks to prove that it is running. Miners will include votes in their blocks. Votes can either be positive (elect not yet elected masternode) or negative (deelect already elected masternode). If masternode has more than 30 positive votes in the last 60 blocks then it is added to the list of elected masternodes. If masternode has more then 30 negative votes in the last 60 blocks then it is removed from the list of elected masternodes. Note that it is possible to determine which masternodes were elected at each particular block.

Ofcourse the vote that is cast should be based on the principle described here, but at the moment high amount MNs get to give their vote  BEFORE the small amount MNs.

That's the problem right there. And THIS IS what has to go.

The problem isn't even so much the maximum 10 votes per block. We could probably live with that, although it should be increased to make everything more dynamic.
But aslong as the high amount MNs get to vote while the rest has to wait, the elections are a farce!

We are not going to have competition this way.

Strumpet!

  • Full Member
  • ***
  • Posts: 158
    • View Profile
Re: Masternodes Testing (Round 2)
« Reply #326 on: February 13, 2015, 08:00:35 pm »
You are getting exactly what you wanted, although the process is slower than anticipated. I told you, there are never going to be any 100 SPR MNs on mainnet. There are very quickly going to be 1440 @ at least 1000 SPR each.

What makes you think we'll have 1440 at 1k each?  That takes up much of the coin supply.

Current coin supply, yes. Three months or six months from now with ~260000 new SPR minted each month... not so much. Forget about MNs @ under 1000 SPR.

Ah, ok.  Your, "quickly" and my, "quickly" were different durations :)

Regardless of whether it's 3 weeks or 3 months, I still think the whole exercise is a waste of time. :) 

IMO Mr Spread should just call it 1000 SPR collateral required and spend his time more profitably on I2P integration, improved UI, secure comms, and other MN services, not some daft and inevitably short-lived notion of 'Anyone can run a Spread Masternode for 5 bucks.'

Strumpet!

  • Full Member
  • ***
  • Posts: 158
    • View Profile
Re: Masternodes Testing (Round 2)
« Reply #327 on: February 13, 2015, 08:01:45 pm »
You are getting exactly what you wanted, although the process is slower than anticipated. I told you, there are never going to be any 100 SPR MNs on mainnet. There are very quickly going to be 1440 @ at least 1000 SPR each.

Bullfrog.

I didn't want high amount MNs to be preferred by the system when it comes to the ability of giving and taking votes.
I don't even know why Mr. Spread decided to use a formula that treats them better than the rest.

WTF?

We are not going to have competition this way.

They are better than the rest. A high collateral makes it economically much harder to attack the network.

georgem

  • Tech Admin
  • ******
  • Posts: 948
    • View Profile
Re: Masternodes Testing (Round 2)
« Reply #328 on: February 13, 2015, 08:06:35 pm »
You are getting exactly what you wanted, although the process is slower than anticipated. I told you, there are never going to be any 100 SPR MNs on mainnet. There are very quickly going to be 1440 @ at least 1000 SPR each.

Bullfrog.

I didn't want high amount MNs to be preferred by the system when it comes to the ability of giving and taking votes.
I don't even know why Mr. Spread decided to use a formula that treats them better than the rest.

WTF?

We are not going to have competition this way.

They are better than the rest. A high collateral makes it economically much harder to attack the network.

That's completely false and upside down!

The higher amount inside your MN doesn't make it "better". It just makes it safer from being kicked out of the list.
What makes your MN "better" should be based on the votes it receives.

LOL!

 8)

How come we planned to create a system that enables competition but ended up with a system that disables competition?

We need to go back to the drawing board, Mr. Spread.

georgem

  • Tech Admin
  • ******
  • Posts: 948
    • View Profile
Re: Masternodes Testing (Round 2)
« Reply #329 on: February 13, 2015, 08:13:52 pm »
Let me give some examples:

It should be possible to install an MN on a mega high powered server with just 100 SPR in it, and have it receive brilliant votes by the network. But it will still be kicked out when it becomes the weakest link...

It's also possible to install an MN on a slow home computer and put 10k SPR there. Now this masternode will constantly receive bad votes, and be probably even kicked out of the system if its performance is totally terrible. But it will be kicked out because it gets de-elected... not because it became the weakest link.