Improving the pool distribution idea

[rentahash]

I posted this on BTCtalk but since devs are busy with V12 release they haven't comment it yet... I am re-posting it here and hope someone from dev team will find time to comment it..

Quote from: rentahash on August 14, 2015, 08:02:11 PM

Quote from: rentahash on August 14, 2015, 05:12:01 PM

Quote from: eduffield on August 14, 2015, 04:25:30 PM
We're definitely open to improving the pool distribution if anyone has ideas. However, PoS is incompatible with anything that uses masternode quorums. The quorums are chosen based off of the proof-of-work hash, so if we were to switch to a proof-of-stake model, we would end up with less secure hashes and other issues that would have to be addressed. That would allow an attacker to pick the masternodes that approve a transaction lock for example, something that we definitely want to avoid.

I might have an idea, but maybe it is too hard to implement.

I believe as we all want to have as much spread of the hashing on one side and pool owners to have much profit from their miners on another side is really hard to meet.

My idea is to use power of MNs in a way that every block which is mined on a pool with the higher/highest share of total hashrate should earn less DASHes then normal 50% of the block. If it can be implemented dynamical sharing the earnings of a block pool/MN we could dynamically solve spreading of hashes as miner will leave top hashrate pools by himself.

So I am see it as like the more blocks you are finding in time the less profit you are getting from it... at the end of the day whole hash spreading should meet the equilibrium of spreading near to p2p pools.

Also pools could compete again in % of fees as they used to...

Maybe it is not possible to implement easily but I think it is a good idea for dynamically change block rewards for miners and MNs.

I really think that this idea has potential, would like to hear from devs if they see it in a same way?
Dynamically block reward is already achieved = difficulty of the network, so we could adjust it more per block with dynamically sharing block reward miner/MN as it is already on protocol level.

pseudo math just to maybe represent better idea:

DBR = dynamically block reward for miner in DASHs
BR = block reward in DASHs = 5 DASH
NSBR = network percentage of block reward for miner in percentage = 50%
BPT = blocks found per time (defined time frame, lets say 1hr) for a miner ,
lets agree maxBPT to be 10 just to prove theory, if miner founds more then 10 blocks per time we will still use 10 in equation

DBR = BR * (1- log(BPT)) * NSBR/100

And for MNs reward is = BR - DBR

something like this

 

[UdjinM6]

The problem is in words "per miner" (pool actually). Right now miners (pools) can be determined because they use the same address to mine to all the time. But once you have this limit in place it's only a matter of (a very short) time when pools will modify their software and mine on new address every block. Nothing stops them from doing that now but there is just no need for them. So imo that way you are not going to achieve anything but you also going to loose ability to "identify" miners as it is now and to have at least more or less accurate overview of their hashrate/block extraction rate.

 

[rentahash]

The problem is in words "per miner" (pool actually). Right now miners (pools) can be determined because they use the same address to mine to all the time. But once you have this limit in place it's only a matter of (a very short) time when pools will modify their software and mine on new address every block. Nothing stops them from doing that now but there is just no need for them. So imo that way you are not going to achieve anything but you also going to loose ability to "identify" miners as it is now and to have at least more or less accurate overview of their hashrate/block extraction rate.

Yes you are absolutely right UdjinM6, but then if you look at chart https://chainz.cryptoid.info/dash/#!extraction

this unknown can also be coinmine for example proxing to another address... that scares me more.

That is why I said it is very hard to implement, but I might have idea how I would do it:

Proposal by indexing all addresses with its hashrates attached to them, once hash is gone from that address you are looking for new address with "similar" hashrates (or if they break on few/many addressees sum it) and group them in clusters... like pattern recognizing movement of the hashes.. so we could still follow equation on a cluster level instead of miner/pool - this should avoid pools to move from address to address as they will likely loose profit by acting funny on the network, as well miners would leave unstable income pools ... Of course that there will be some issues and no pattern recognition software is perfect... but I am sure it could reach over 95% accuracy soon... remember that normal miners who are not playing around will not be affected with this one.

problem is of course that this is open source and everybody will see the algorithm of pattern recognizing, and trying to act out of scope... I am aware of it, but there are smart people in our community, I am sure it could be found bullet proof..

BTW thank you for your time reviewing idea!

 

[UdjinM6]

Yes you are absolutely right UdjinM6, but then if you look at chart https://chainz.cryptoid.info/dash/#!extraction

this unknown can also be coinmine for example proxing to another address... that scares me more.

That is why I said it is very hard to implement, but I might have idea how I would do it:

Proposal by indexing all addresses with its hashrates attached to them, once hash is gone from that address you are looking for new address with "similar" hashrates (or if they break on few/many addressees sum it) and group them in clusters... like pattern recognizing movement of the hashes.. so we could still follow equation on a cluster level instead of miner/pool - this should avoid pools to move from address to address as they will likely loose profit by acting funny on the network, as well miners would leave unstable income pools ... Of course that there will be some issues and no pattern recognition software is perfect... but I am sure it could reach over 95% accuracy soon... remember that normal miners who are not playing around will not be affected with this one.

problem is of course that this is open source and everybody will see the algorithm of pattern recognizing, and trying to act out of scope... I am aware of it, but there are smart people in our community, I am sure it could be found bullet proof..

BTW thank you for your time reviewing idea!

Hashrate is what only pool itself knows - it's basically "how many hashes on average I have from my miners per second". Block hash is one and only (single) hash out of that huge number of hashes that pool got from its miners. There is no pool hashrate in blockchain, only a single hash that matches current difficulty. And so, there is no hashrate "moving" too.

 

[rentahash]

Hashrate is what only pool itself knows - it's basically "how many hashes on average I have from my miners per second". Block hash is one and only (single) hash out of that huge number of hashes that pool got from its miners. There is no pool hashrate in blockchain, only a single hash that matches current difficulty. And so, there is no hashrate "moving" too.

I see, but then I dont understand how https://chainz.cryptoid.info/dash/#!extraction is getting this chart with hashrates?
Sorry for this question, it seems that I misunderstood this chart then

EDIT: is it block found piechart then?

 

[UdjinM6]

I see, but then I dont understand how https://chainz.cryptoid.info/dash/#!extraction is getting this chart with hashrates?
Sorry for this question, it seems that I misunderstood this chart then

EDIT: is it block found piechart then?

When miner/pool mines a block he create a coinbase transaction where output(s) is his address(es). That's how miner/pool get it's reward and while he always use the same address you can recognize him. Then you just count number of blocks with such known coinbase outputs to get chart like this.

 

[rentahash]

When miner/pool mines a block he create a coinbase transaction where output(s) is his address(es). That's how miner/pool get it's reward and while he always use the same address you can recognize him. Then you just count number of blocks with such known coinbase outputs to get chart like this.

thank you for clarification Udjin

 

[Paws]

By that logic, wouldn't you be able to hire cloud mining for a day or so to your address and suddenly you would be penalized for power you purchased? Sort of like buying a lottery ticket for $5, winning $10 and being taxed $5 for winning by the tickets rules?

There is money to be made in buying hashing power when trading out other alt coins, but how it's done is oblivious to me, this hash renting would be useless with an address penalty, especially since bulk hash power at fixed rates is more expensive rather than less and if I understand it correctly, renting miners hash power could be a good thing since you move that power off of a centralized website, and into your own pool, albeit temporarily.