Not quite... If pool hashrate is fairly constant but pool luck is not - ie, the time to find each block varies quite a bit - then any given miner will submit way more than n shares for a block that took much longer than average to find; ergo, the oldest shares end up get tossed out because they were outside of the n-wide window.
I see, that explains my pathetic returns while hashing out on Neoscrypt. When there is too little pool wide hashrate in the algo, people who keep on that one algo suffer. I quit because of those returns, and looks like it was a good decision.
And I am not complaining about the first couple of days, there was plenty of hashrate and got paid fine.
- the downsides to PPLNS are basically the result of trying to apply a mechanical rule - payout based on the last N-shares leading up to each block - on a statistical process - the time to find a block does vary in inverse proportion to pool/network hashrate, but it is not a linear function; that is, as pool/network hashrate declines the odds of finding a block decrease much more
rapidly chaotically (but asymptotically - it never quite reaches zero for a non-zero hashrate).
What many miners understand is that if pool hashrate for an algo declines while network hashrate remains the same - e.g., most miners switched away from that algo on a specific pool - then pool
luck tends to decline as well. Intuitively one may think that as pool hashrate declines it will take longer to find each block, sure, but each share will be proportionally more valuable - lower hashrate, right - so overall payout should remain the same. However, what really happens is that a lower pool hashrate not only increases the time to find each block
on average, it also generally results in a
wider variation in the time to find as well. THAT is what gets you with PPLNS: the number of shares considered valid for each block remains the same, but it is much less predictable when each block will come, and with PPLNS you only get paid for shares leading up to a block.
Pool operators like PPLNS because they don't have to pay for some of the shares submitted if the time to find the next new block suddenly jumps up as a result of an equally sudden drop in pool/algo hashrate. Miners also eventually learn to remain on the pool/algo until a new block is found to maximize their payout - hence why PPLNS is said to reduce pool hopping, because you can't be sure you'll receive anything if you drop in on a pool/algo for a few minutes or hours without a block being found in that time.
What people wrongly assume - and which few pool operators attempt to correct - is that miners will almost certainly be
persistently underpaid if the time to find is "too long" at a given PPLNS pool (where "too long" is a nebulous, hard-to-pin-down number, unfortunately). This is one of the reasons why hashrate tends to concentrate at certain pools for certain coins/algos; it's a rather perverse example of the "network effect" showing up in something that is supposed to be decentralized/democratic by nature (ie - cryptocurrencies).
NB - I am fairly new to mining so my understanding of all this may be imperfect, incomplete or even incorrect. I welcome being corrected; I do not respond well to being called names.
