I believe this is the address associated with this miner:
https://explorer.zcha.in/accounts/t1KHa9CJeCy3b9rUX2BhqkFJXSxSSrhM7LJ
This comes from looking at recently mined blocks on coinmine pool and comparing the frequency of associated addresses. This is the address associated with blocks mined by “anonymous” users that occurs the most in the past few hours.
He has sent/received more than 10% of the entire available ZEC supply.
The source of this hash rate looked pretty obvious to me, rate drops to 671,024 Sol just as an order on Nicehash drops to 670 K/sol. At a rate of 0.0132 BTC per k/sol a day you only need to pay 0.37 BTC for an hour of that. 
The white paper and equihash algo implemented in zcash mention using algorithm binding. You have to use Wagner algo to mine a zcash block.
This account is related Accounts - Zchain
The balance is 9306 ZEC. With an available supply of 46 455 ZEC, it’s exactly 20%.
Founders Reward ?
All of the founder’s reward multisig addresses can be found here:
Ok, I understand better. Thanks
binded to maintain consistency on wagner. Not true for all.
“Algorithm binding: The generalized birthday problem
in its basic form lacks some necessary properties as a
proof-of-work.”
What do you mean by consistency? Where is that stated in the whitepaper? The algorithm binding is what makes equihash a supposed valid proof of work. In order to be a valid solution you have to pre-fix the positions where 2l-XORs have zero bits, thus we bind the user to a particular algorithm flow.
The question is whether or not the propositions about the algorithm binding ensuring a memory hard pow are true or not.
Comprehending this mathematical thesis needs better of me. 
if it was a single farm it could have mined solo!
it should be millions of raw cpu power throughout the globe!
maybe his solutions are fake and he just scam pool to get reward, when with solo he would not do any block
Looking at the rising hashpower, my guess is that he is deploying miners to a botnet.
There is some optimization headroom for miners, but I do not think that is could be this much. 
That is one possible theory, but remember that botnet owners do not get to be picky about the CPUs they use. Assuming 10 H/s per CPU - which is very much on the high side - this is a botnet of 300 000 computers or more. Almost certainly many, many more.
Now botnets that large are indeed possible, but remember that each compromised machine is a separate worker, and coinmine’s servers supporting half a million simultaneous workers on one account seems more far fetched.
yup. so, either he has a very optimized, super hashing cpu solver or a very optimized gpu miner.
either way, to those who say he could have mined solo: he could also setup hiw own pool which should not take too much.
It’s not a shift in network hashrate from Ethereum which means these are probably not GPUs. It would’ve already been present under one of the algos.
Heavily botneted algos like Cryptonote haven’t seen a huge drop in difficulty, so this is new.
Guess this is AWS instances on highly optimized CPU miners. Someone figured out how to make that profitable, that’s why it changes so much and came out of thin air. Looks like it scales up and down with network profitability.
Goes to show you there is a lot of leeway that can still be made with GPU miners if CPUs are still competitive (even on discounted amazon credits).
So its not the massive orders on the nicehash site?
Hmm. They do have around 3.5M sol/s available, by the looks of things, which means that maybe the numbers do come close. But I don’t think it would be possible to hire all of them at one time.
It is, but it seems that the people here like to believe in conspiracy theories about some giant botnet.
Ummm… probably not conspiracy but possibly. There are multiple large botnets for sale right now on the deepweb; where one could easily run a low resource miner… over 100,000 nodes would equate to some hash power.
That’s not necessarily the case here; maybe Amazon or another large scale cluster…
Hiring all of them at once is difficult.