09/19: Proposal of block reward adjustment to depress difficultyInitial designWe have to go back to review the initial design.
https://bt.irlbtc.com/view/735170.msg8755719#msg8755719
Predicted consequence is fair distribution, avoiding big mining farms and maintaining the network hash under a certain level (unlike BTC network keep increasing). It will frustrate mining farms (if existed) when network hashrate is so high that the mining rewarding is at the decline side. Mining XMG is less profitable compared to mining others; only belivers keeping mining will receive the most out of it. Same frustration applies to GPU miners if they existed.
According the above image, sorry for big miners, my initial thinking is to exclude the (personal) big hash as much as possible. Since we are initially based on M7, GPU mining is allowed at that time. The situation I am trying to avoid is the GPU farm, that is the reason of using diff-dependent rewards. As to M7M, allowing big hash in network means allowing GPU miners too if they exist. I do believe CPU miners (even with farms) would hate to see GPU farms online, same applied to individual CPU miners. This initial direction must be complied. IssuesNetwork hash keeps increasing; we haven't see the arrival of rewarding peak yet while we already noticed the coming of big miners; Current mining status in two pools:
Suprnova Workers: 244
Nonce-Pool Active Workers: 183
total works: 427
Total network hash is about 105 MHps.
It would be conclusive that individual miners have reached saturation, further increase in hash is mostly due to big miners. We haven't got into the 500 XMG/block stage. Provided the general user base won't grow (most likely), what do we expect once net hash hitting a point where gets 500 XMG/block?
Solutions1) We intend to get 500 XMG/block into the net, but conditional
2) This is the idea; let's say we have three mining groups
Group A - CPU individual miners
Group B - CPU farms
Group C - GPU miners & farms
Though we may expect nonexisting of GPU miners, we can't exclude the possibility (individual). This must be taken care of.
The most importance of this solution is to maintain the #A:
a. Mining hash of #A must remain the same or above during the lifetime of mining.
b. Moving the optimum diff to a point where the net hash equals to the hash of #A; optimum diff is where we get maximum rewards.
It will be clear that with the arrival of #B, #A+#B leads to less rewards; Both #A and #B suffers low rewarding. #A is a faith miner and never go way in spite of zero reward or 500 XMG, and what happened to #B. #A is a set of many individual miners; for one who mines with 1 cpu, it won't be much loss if nothing coming out in one or two days (in an extreme case, we can assume at #A+#B rewards is zero). But for #B, since #B is a big farm and running costly, it won't be acceptable for them to mine without any return. We can image the same thing for #A+#B+#C. Obviously, the only way of getting rewards for #B or #C is to lower their hashing power, so that the block rewards are available for mining; it may get them to the same hashing level as #A.
If we can go this way, we will will need to select optimum diff and reward-diff pattern. The standard can rely on the mining history so far.
Any suggestions are appreciated!
Please suggest if this is the way we can go.
09/21: Random IRC talk regarding the new block rewarding systemhttp://cryptomagic.com/files/misc/irc-log-09-21.txt7:47 PM <@joelao95> https://bt.irlbtc.com/view/735170.msg8919002#msg8919002
7:47 PM <@joelao95> check this post about block rewards
7:47 PM <@joelao95> all please hold one
7:47 PM <@joelao95> *hold on
7:48 PM <@joelao95> let's push all big miners out of the scene
7:48 PM <@joelao95> then all of you will get rewarded with 300 XMG/block
7:49 PM <@joelao95> bojcha, how many coins you got before and now
7:50 PM <@joelao95> yes, the optimum has is calculated based on individual miners
7:51 PM <@joelao95> we supposed to get 300XMG at 40 MH/s
7:51 PM <@joelao95> hi paul
7:51 PM <@joelao95> let's say one cpu get 50 kh/s - 100 kh/s
7:52 PM <@joelao95> no problem, that's fine
7:52 PM <@joelao95> so 40 MH/s means 400 - 800 cpus
7:53 PM <@joelao95> if individual miners remain 40MHs, there is no miners with higher hash or gpu miners will get profits
7:54 PM <@joelao95> the key is that individual miners must keep mining
7:54 PM <@joelao95> paul, yea, that's the issue too
7:54 PM <@joelao95> bojcha 
haha
7:55 PM <@joelao95> ok, let me ask you
7:56 PM <@joelao95> if we have 300 XMG/block, and the reward won't reduce at higher hash; regarding the situation we had, there are few miners with 10Mh/s above, would that make you stop mining
7:57 PM <@joelao95> even worse, we've see network hash keep rising, and there is a concern of gpu miners, we could see many more 10Mh/s miners
7:57 PM <@joelao95> will that make you stop mining
7:57 PM <@joelao95> we have no chance at all in that situation
7:58 PM <@joelao95> am I right, or missed something, correct me
7:59 PM <@joelao95> paul, the issue is that situation just make things bad
8:00 PM <@joelao95> if someone has 100 vps
8:00 PM <@joelao95> there will be others have 200 vps, 300 vps etc,
8:00 PM <@joelao95> never stop
8:00 PM <@joelao95> BTC is an example
8:01 PM <@joelao95> ok this is the situation with small miners i would expect, say each person can only mine with 50kh/s , there will many many people join in the mining part
8:02 PM <@joelao95> but, if there are few 10MHs, could be like this, 1000miners + 10MHs miners
8:02 PM <@joelao95> what if there is a 1GHs miner, could be like this 100 miners + 1 GHs miner
8:03 PM <@joelao95> not like that; but I rather put limitation for one who just keep increasing his power hash
8:04 PM <@joelao95> let say you rent two vps, you will get twice pay
8:04 PM <@joelao95> three vps, three times
8:04 PM <@joelao95> but you rent 1000 vps, then, won't be 1000 times pay
8:05 PM <@joelao95> there is a point, go beyond that people will get loss
8:06 PM <@joelao95> yea, so this is risky
8:06 PM <@joelao95> there is price to maintain this
8:07 PM <@joelao95> paul, the situation is if we don't put this limitation, I suspect there is gpu miners soon,
8:07 PM <@joelao95> XMG will die soon or later
8:07 PM <@joelao95> at least the current situation (few 10Mh/s) already freak out miners
8:08 PM <@joelao95> I don't think of any new algo could really resist gpu
8:08 PM <@joelao95> though I don't M7M
8:08 PM <@joelao95> no confirmation
8:09 PM <@joelao95> but new algo just create hidden miners people suspect GPU miners
8:09 PM <@joelao95> that's a FUD
8:10 PM <@joelao95> this is pain
8:10 PM <@joelao95> any pain we will go though this stage and then enter PoS
8:11 PM <@joelao95> *anyway
8:11 PM <@joelao95> I showed this idea to Hix and itod
8:12 PM <@joelao95> they though this quite innovative, though there are unknown things. but deserve try out
8:16 PM <@joelao95> will closely keep an eye on the nethash and block rewards
8:22 PM <@joelao95> anyway, paul, this type of block rewards is designed from the beginning, not just because I've seen the situation now and changed my mined; what changed now is better to fit this algo
09/21: MagiPub: the new block rewarding system explained - A network-dependent rewarding system: proof-of-mining1. MagiPublicationI've managed our first publication in the academic level, currently submitted to arXiv.org e-prints:
A network-dependent rewarding system: proof-of-miningRaising up this effort has dual purposes. We've experimented a few new features in Magi in my expectation at this coin going towards a direction never achieved or attempted before. We are kinda researching something and such efforts should be recorded and published if possible. XMG does benefit a lot from either suspect & complaining voices or from positive comments in the thread. With many back and forth discussions, we seem like going toward a solid direction now. If XMG cannot achieve whatever it's supposed to be, subsequent coins might go stepping ahead of us by benefiting from the publications.
I, myself, was working in the research field (credit: RG score more than 29 based on ResearchGate). I like to think and ask. I have been kinda addicted to the cryptocurrency since last Dec, and have thought about a lot over the time. I don't want point out for whatever reason I am doing this, this coin is just on the track, and I am not just joking in making a random coin, but rather seriously take it in the long run. I won't object to the argument of pump and dump (just meaningless wording arguing). If I did wrong the coin, correct me; I am very like to hear different voices. Back to the topic, publishing papers on XMG will obviously place XMG into a more broader audience, just imaging how many people will read the publication.
2. New block rewards explainedWe were missing a full description of the new rewarding system:
A network-dependent rewarding system: proof-of-mining.
(a) - our initial design;
(b) - the actual implementation in the M7M algo. Ads. (anyone can get a better ads or image based on this idea, shoot me a PM)
1000 people mining BTC/LTC with their CPU (assume 50Kh/s) only get negligible rewards; however, mining XMG yields 144 coin/day/person.Current mining situationMiners should be aware of the current low block rewards (5 XMG/block typically) is because of the overtuning of the rewarding system; that is, we are at III of figure (b), rather than I. This is important; big miners might need to lower their hashrate a little bit, so that the reward will be back to maximum (at I). Deep explanationWe have explained the rising part in the initial design: the system allocates more rewards to the network having more miners. However, simply increasing reward will drive miners towards building more powerful equipment. The relationship between mining costs and returns in consistence with economics should be the guideline to achieve the goal. We assume this is valid. For example, one who has powerful equipment and hence high cost as well will expect commensurate returns; otherwise, the cost must be reduced by reducing the usage of equipment.
The efficiency of discouraging mining activities is fully determined how fast the rewards decline with difficulty (at the high difficulty range). It should be noted that self-adjustment of network difficulty in accordance with rewards is affected by both the reward declining rate and the market place of the cryptocurrency. One can build up more powerful equipment to compensate the declining rewards. A much faster declining rate will disable any possibility of acquiring desired rewards by simply building more hashing power. Fig. (b) shows an improved rewarding system as implemented in XMG, by incorporating a cut-off function. The maximum block reward is available at Diff = 1.75 (I) (or ∼ 40 MHash/s). The reward is halved when Diff = 2.20 (or ∼ 51 MHash/s) (II), and equals to 1/10 of the maximum value at Diff = 2.37 (or ∼ 55 MHash/s), at which the low rewards will be lack of incentive to the miners.
Does the new rewarding system work?The short answer is yes. Let's take a look at the pool status before and after the hard fork (new reward):
BEFORE (nonce-pool and suprnova, respectively)
AFTER
Summary| ==================== | ==================== | ==================== |
| Network hash ratio | big miner weight (>5 MH/s) |
| Before (at block=2660) | 115 MH/s | 47.5 % |
| After (at block=5682) | 83 MH/s | 38.4 % |
| ==================== | ==================== | ==================== |
The big miners did have been depressed to a certain level; we will carry out proof-of-mining campaign to help this further.
Proof-of-miningOne may expect the market place will affect the miners behavior significantly. For example, the high-pricing reward leads more miners into the network which immediately move the block reward out of the maximum region. This initiates a self-adjustment of the rewarding system. Realizing the marginal rewards (e.g., at III of Fig. (b)), some of the miners start ceasing their minings, thus pushing the reward back to the maximum. In this case, it will be relatively less advantageous for miners quickly in and out the mining activity. One will need to show a proof of mining in order to gain sufficient credits to declare the rewards. The new system emphasizes the participation of miners instead of their hashing power.
Another look - difficulty vs time
It's clear that diff is nearly constant between 09/18 and 09/21, while it declines after 09/21 when hard fork happened.
Currently over 530 active miners.
