Topic: ckpool.org CLOSED - page 89. (Read 162894 times)

When the pool was 10ph or less  blocks came slowly  and our luck was decent.

So pressure of variance simply did not happen.

Now at 168ph  we should hit  5 blocks in a week

So if we hit 2 blocks  we are short 60%

Ie 0.0100 in payouts is 0.0400  that is stressful

We hit  these last 5 blocks  and my payouts  went from .041 to .023  due to high hasher stripping  my %  = bad for me

We hit these last 5 blocks  and we should have hit 9 with average luck  = bad for me.

All of this was has happened in about 35 days  which makes it seem worse.

First 8 months we did about 8 blocks and had average luck.

Last month we did about 5 blocks and had shit luck  

In my case  I collected around

0.1500 with poor luck

if luck was "normal"

it would have been

0.2700 btc

but  the shortfall is true for all of us.

Now I have  t1's here and they do not have much choice  but I could have them at f2pool  doing pps I did not do this.  My decision  to mine here will be me mining here and I won't grumble about "bad" luck

I always liked -ck  so for now I am here.

It just past 12 noon  EST  where is that block?

ViaBTC has historically been a classic 4% PPS pool.  They dropped that to 2% for a few months after an issue as a "sorry" response.  If they are down to 1% now its likely because they are keeping a part of transaction fees.  

3.5% for BTC.COM as a PPS pool sharing full TX fees is not a bad deal - IF your need PPS's lower volatility.  

However, unless your a smaller miner (say a S9 or two or less), PPS will cost you more over PPLNS.  Its just a question of risk management (aka variance acceptance).  Slushpool at 2% is large enough you wouldn't feel that variance if measured in terms of months not days.  Kano at 0.9% is good if measured over quarters not weeks.  Ckpool is smaller than Kano, so has even larger variance, but has grown enough recently that your very likely going to find several blocks per difficulty period - which is big enough for those with a very long time horizon.

For those smaller miners "dust" becomes an issue - where your reward is less than the fee associated with sending it to you.  But even there, just select a pool like Slushpool that allows you to set a payout thresehold so your dust can accumulate to something reasonable before payout.

For the 1st 10 blocks I was at 1.14-0.25; if we hit right this second, I'm at 0.09444954 (based on a 13 reward).



1) the "now" is your answer to that one. Screwing people early on and paying "more" a year later is a PR stunt that anyone can do.
2) via charges 1% on solo and up to 4% on PPS (with PPS paying "up to 96.91%" of expected)
3) How does one use that nickname and not do any actual math before posting?

[/quote]
8 years of pooled mining and input from many maths geniuses has failed to find a "better way" than PPLNS - the stochastic nature of block finding is the actual issue. Even if I charged 10% fee and ran PPS the pool would have gone bankrupt by now. A different cryptocurrency with a deterministic block finding would be so much easier, but so much more game'able.
[/quote]

Respectfully...that argument does not seem to hold true...Via & BTC.com should be evidence enough no?
(Via is currently 1% and pays a portion of the TX while BTC. is 3.5% and shares full TX now)
If it is as you say...How do they survive and why are they the top two dogs in the race?

(And I and Tesla would argue those Math experts are idiots and have no relation to true genius..Meni Rosen schmuck especially)

even noon HST  

The Hawaii–Aleutian Time Zone observes Hawaii–Aleutian Standard Time

I am down to .023 a block

or even noon PST would be good 

Let’s get a quick block before noon eastern standard time.

13+BTC and confirmed!

Party!!

https://blockchain.info/block/0000000000000000001fe99efc67100475ffe141d735e508de2b69c0df6347db
Block!

Understood. As you know all the code for this pool is public.

Accounting is done in this function here:
https://bitbucket.org/ckolivas/ckpool-splns/src/13ec0efc5089f0d8f4ef658c8618663634d40648/src/stratifier.c?at=splns&fileviewer=file-view-default#stratifier.c-6113

Pool stats are added under lock here:
https://bitbucket.org/ckolivas/ckpool-splns/src/13ec0efc5089f0d8f4ef658c8618663634d40648/src/stratifier.c?at=splns&fileviewer=file-view-default#stratifier.c-6137

User critical stats (herp) are added under lock here (it even says so in the comment):
https://bitbucket.org/ckolivas/ckpool-splns/src/13ec0efc5089f0d8f4ef658c8618663634d40648/src/stratifier.c?at=splns&fileviewer=file-view-default#stratifier.c-6150

However the client, worker and user hashrates are calculated here (decay_client, worker and user):
https://bitbucket.org/ckolivas/ckpool-splns/src/13ec0efc5089f0d8f4ef658c8618663634d40648/src/stratifier.c?at=splns&fileviewer=file-view-default#stratifier.c-6168
As they're not done under lock and multiple threads receive shares, there is a race on the stats. With 3200 shares per second that one worker and user is often overlapping and the race is ruining the stats. The decay_ functions use expensive maths functions so locking for that duration was originally considered to hold up threads too much. I never anticipated one user with more than 3000 workers though...

If anyone below me (and I say "below" simply because I can't quite compete with owning 5+PH and/or research/dev costs) on the payout list wants to compare cost vs revenue (understanding that most of my rented hash was purchased with a $9k+ BTC and I've, literally, never been under 2% of the reward until last week), we can talk math all day.  

dont fuck with stats.

the pool did fine  for the first  8 of the 9 months

the last month sucks

so to say we took a beating for 9 months  is a lie.

to say we have been hurt this last month is true.

To say We have been crushed  since we expanded hash rate is true.

And 5 1% blocks in a row makes up for the  cold streak


I trust -ck  but my complaint is this.

It looks bad.  I know  enough about programming to know  -ck 's  explanation is possible and I continue to mine here, but it looks bad.

It matches the bad luck.

8 years of pooled mining and input from many maths geniuses has failed to find a "better way" than PPLNS - the stochastic nature of block finding is the actual issue. Even if I charged 10% fee and ran PPS the pool would have gone bankrupt by now. A different cryptocurrency with a deterministic block finding would be so much easier, but so much more game'able.

"Your anger is misguided"
No it is directed at myself alone
I have nothing but respect and admiration appreciation for CK's expertise and efforts. And many of his opinions regarding centralization.
(And to a lesser extent Kano's, even if he is an arrogant asshole)
I also had complete trust and confidence when I recently sent +10K CDN to buy some (NON BTC) machines from CK.


"Presumably the catalyst for your outbreak was the latest fall in bitcoin value"
No it is a complaint against the PPLNS method and the terrible run of luck we had.
Made worse by the influx of a Mega-Whale.
While the risk is sugar coated by the mantra don't worry it will average out in the long term. It is a seductive fallacy
Which explains why we don't complain when we hit early lucky blocks..because we SHOULD have some of those.

"Do you see it as a public service to prevent other people from losing their money by mining here?"
..."you are worsening the situation by making more people try and move away"
No, The mass are idiots and I have no doubt few will read and even less will care or be influenced.
Walter Russell remains obscure overlooked and underappreciated despite his staggeringly important achievements in science which by far eclipse those of his friend Tesla .
Why would I care what others do? It would serve me better that they suffer!

"Do you think I run this pool out of malice to make miners intentionally lose money?"
Certainly NOT and I do wish apologize for inferring so..That was rather childish...I am embarrassed and wish I had not lashed out so.


"but there is absolutely nothing we can do about it"
I disagree, PPLNS is the problem...I can't accept that there are NOT other better methods which can protect the operator AND greatly reduce the risk to miners. Irrespective of pool size.

I SHOULD HAVE made a more polite and elegant argument and appeal to consider changing to a PPS system.
I for one would be happy to pay 4/5% to CK to code and operate a pool to aid in decentralization
I would stay and pay an even higher % if the fee was variable and linked proportionally to pool luck and Hash rate.

So I find his arguments re: providing a service and concern of centralization somewhat hollow.

I will grant that it would be a major pain in the ass and would respect his choice not to take on such a large responsibility.
Certainly I would not expect it to be free.
Nor do I accept that he is not competent enough to do so.

"you've finally had enough of mining at this small pool"
NOT true, it is my proffered choice, I just can't afford the risk.

"why people continue to flock to larger pools - because of the lesser variance the large pools provide"
Not true..."Post hoc ergo promter hoc"... fallacy...
NOT true..Large Pools...DO NOT have LESS variance..Nor will the faster block times affect variance...
THE MAJOR REASON IS THE PAYOUT/REWARD SYSTEM.

with 3 EH as opposed to 150PH
because math

To be fair, slush show block luck the opposite to how ckpool does, so lucky blocks have a higher pool % of luck.

Look at the time taken for those two blocks, they are only 34min and 10min blocks.

Thanks CK and CompGen. We'll have a block party today I'm sure.   

Actually, it does. I'm far too lazy to try to explain the math; however, the simple fact is that when bad luck hits, we all lose far less at 200% on a run of 2 day blocks than we do at 110% on a run of 35 day blocks.

People tend to err on the side of confirmation bias in times of less than optimal results. As you can see by the image I posted earlier, shit happens and it happens to every one. Take the heavy hitter burn-in out of the equation and if we were to hit the same bad luck as slush, we'd hit 2 blocks in the next 5.5 months and both would be in excess of 5N (which would mean that 5N would actually only pay the expected of 1N - just like it did at slush).

Shit luck is shit luck and no pool is (or ever will be) immune from it.

It may be a "Halong pool" because most of the hashrate is from Halong but it's not a Halong pool in reality because I'm the one choosing the transactions and work to include, and I use default bitcoin core daemon rules. In that regard, it's totally different to bitmain miners being owned by bitmain mining at a pool run by bitmain.