Last year or so luck has been brutal? Luck for the last year (250 Block Luck) is 92.35% which is not too shabby no matter how you look at it. It's well within the expected range for that time frame.
It seems proclamations about the death of the KanoPool have been greatly exaggerated.
The problem is other people having no understanding of statistics and making claims about abilities to predict the future.
Seriously, that is actually what happens.
The 250 Block Luck (still) includes the two worst blocks of all time in Oct 2017 - the 928% and 875% blocks.
They are currently blocks 234 and 236 counting back.
Now if we calculate the worst run of 5 blocks around those two (that is also faking statistics to get a worse results) we get:
2166 491758 12.99319213 2017‑10‑26 00:42:04 Matured 11,115,667,052,925 928.788% 1.000 82.01%
2165 491121 13.64023624 2017‑10‑22 09:10:06 Matured 638,937,484,942 53.387% 0.414 88.48%
2164 491091 13.55697548 2017‑10‑22 05:11:09 Matured 93,569,914,525 7.818% 0.075 87.80%
2163 491085 13.00374345 2017‑10‑22 04:35:54 Matured 10,474,247,889,141 875.193% 1.000 86.71%
2162 490351 13.79950370 2017‑10‑17 19:13:11 Matured 849,926,190,745 71.017% 0.508 91.16%
Or a total of 1936.203% for 5 blocks = a mean of 387.2406% per block.
The CDF[Erland] of that is 0.999972 or about 1 in 35714
So yep that's pretty unexpected if we rig the numbers to be as bad as possible.
However, that occurred in the last 250 blocks, so any statistics that includes those two blocks will show low for quite a long time,
irrelevant of how the pool has performed since then.
Luck does NOT always magically reverse itself.
How is luck expected to approach 100% after a large fall in luck?
Simply by the fact that if you continue to get the expected 100% luck, lower luck will approach 100% over time.
The currently shown, 108 blocks on the web site, is effectively showing the last 207 blocks in the "100 Luck%" statistics.
That shows the number (as I already stated) going up and down from 91% to 108% and again I discussed that above.
The luck since then is not unexpected, it has been going up and down, just the pools size has been small so that the variance is higher.
So what's the point of this?
Firstly, no, people can't predict the future.
Yes bad luck happens.
Since then, the stats have actually been within expectation.
Expecting that historical luck will match future luck is not what you can do on this pool.
I regularly run detailed statistical analysis of the pool - well above and beyond what any other pool does.
So that we can identify such problems if possible and act on them.
You can't ever expect good luck on any pool to mean future good luck.
You
may be able to expect bad luck on any pool to mean future bad luck, if they don't do a high level statistical analysis of the pool data.
Which is the case for most pools - and some do absolutely none at all.
Now, what happened after that issue in Oct 2017? I explained this before but will repeat it.
I added a bunch of new code to KDB to verify all shares coming into the pool that they matched the results that ckpool got for them and also checked for blocks and submitted the blocks from KDB also.
Effectively a double check of everything with completely independent code, independent of the less accurate calculations done in ckpool.
Then I reran KDB with the 3 months of data around the time of Oct 2017 on a backup server, and verified all that data a seconds time.
Results: no missing blocks and all shares came out correct.
So the only 2 possible assumptions I can make are:
1) It was just a run of very bad luck
2) There was a large source of hash rate that caused (accidentally or purposefully) some of the loss of blocks
At the time, about half the pool hash rate (for a short period) came from NiceHash
So that was the only hash rate on the pool that
could directly cause such a problem, if it wasn't just bad luck, or only partially bad luck.
The other large hash rates on the pool, large enough to 'possibly' cause it, were known miners who I can assure you did not cause it.
The probability of causing a "16 block withhold" in 10 days with
anything but a large hash rate, is obviously way too low to consider.
Thus I have since then enforced a ban/block on rental hashes, since, as I have stated on the forum many times since then (and before then), there is a financial incentive for rental hashes to withhold blocks, and that risk is too high.
.
.
Now that all leads to another point about all this, and that is about slush.
Why do I bring them up?
Because they are a very large pool who has had worse than this happen on 2 occasions, and yet I don't see reminders about the specifically known withholding fact that did occur there in Dec-2015 and slush effectively hushed up and made all his miners pay for it, or the more recent case of possible withholding where they had a run of 5 blocks worse than our run of bad luck in Oct 2017, yet no one seems to even care about it and nothing was done about it, it was just brushed under the carpet.
i.e. my point being 2 things:
1) that bad luck does occur on all pools, variance makes it more obvious on pools our size, but the large pools that most people mine on don't do anything useful about about (slush did nothing about it for over 5 years and then initially ignored the problem when others pointed it out)
2) you certainly can't know if it is bad luck, or some other problems without a good analysis going on. Personal opinion counts for nothing at all.
I really don't care that Phil doesn't mine here, it doesn't bother me all if not everyone mines here.
Unfortunately pool size does make variance much more obvious, but Phil's withholding statement has absolutely no basis, it's just actually an incorrect opinion based on some emotional outburst, and nothing more.
I left about a year ago. When I left I had averaged about 102% luck. The two years + I mined here.
The last year or so luck has been brutal.
I suspect the pool has suffered with holding tactics on more then one occasion.
A large wealthy farm can do that to smaller pools.
I do believe it has happened here. Which is why pool has had long dry speeds of luck.
Prior to this pool I mined at bitminter and I think he was attacked by withholders.
I also mined at btc guild which was attacked by witholders.
Specifically
I suspect the pool has suffered with holding tactics on more then one occasion.
A large wealthy farm can do that to smaller pools.
I do believe it has happened here. Which is why pool has had long dry speeds of luck.
This is rubbish, it has no statistical backing at all, it's either emotional, or more likely some form of bias.
I don't really care what he 'suspects' coz it's not based on anything other than uninformed guesses.
There's a large bunch of terms and examples related to this
Confirmation Bias, Frequency Illusion, A Reassuring Lie, Gambler’s Fallacy, Anchoring Effect, Availability Heuristic
What they help to show us all, is that people in general have all sorts of reasons for bias and ignoring statistics, but conversely it is statistics that we should take notice of.
