Can an SMPPS (Shared Maximum Pay Per Share) pool be hopped?

I've looked at the way SMPPS (Shared Maximum Pay Per Share) payout schemes work, and as far as I can tell, there is no significant way to hop such a pool to advantage. Yet I've seen several people claim that SMPPS pools should not be considered hopper-proof.

Is this true even if the pool owner seeds the pool with a decent starting reserve of funds and the reserve never hits zero? Or is this only true if the pool reaches the point where it cannot make full payments to miners?

In other words, is this a realistic fear, like it is with PPS (Pay Per Share) pools? Or is this just something that comes up in a very unlikely scenario with a very unlucky pool?

David Schwartz

Posted 2011-10-13T18:21:43.140

Reputation: 46 931

1Probably best to explain abbreviations on first use so as to make posts usable to the internets at large. – eMansipater – 2011-10-13T19:32:49.760

Answers

SMPPS can be hopped, but not in the same way that proportional can. Proportional is hoppable because the expected reward per share is different at different times, so hoppers mine when the reward is high (early in a round) and leave when it is low.

In SMPPS, the expected reward is in theory constant, but the maturity time, the time it takes to actually receive the reward, is not (maturity time is bad because of the time value of money, uncertainty wrt the future of the pool, etc). When the buffer is positive, SMPPS is very attractive, as it is like PPS but with 0 fee. When proportional pools become extinct, hoppers will choose to mine in SMPPS when its buffer is positive because it is better than alternatives. When the buffer is negative the maturity time is high, so the hoppers will leave, and the honest miners will suffer from more than their fair share of the bad times.

So, the hoppability relies on the possibility of a negative buffer. But this possibility is very real - the underlying stochastic process is similar to the one in PPS, but the fee which helps PPS stay afloat is absent. This means that with probability 100% it will become negative at some point, no matter what the initial buffer was.

See also Analysis of Bitcoin pooled mining reward systems for more information about reward systems.

Meni Rosenfeld

Posted 2011-10-13T18:21:43.140

Reputation: 18 542

I'm stuck in newbie land and can't comment on David Perry's answer. I don't know what the proper etiquette is so I'll comment about it here. PPLNS and other score-based methods do not attempt to detect hoppers and penalize them. They do not have a disadvantage (other than slightly higher variance) to intermittent miners. They do not hurt hoppers either, they just give on average a fair payout to everyone no matter how they mine. SMPPS, though, is hoppable to some extent as I've explained in my answer. – Meni Rosenfeld – 2011-10-16T18:56:05.487

SMPPS (and variants) can only be hopped to the extent that one can get FASTER payouts but not MORE payouts. A pool with large backlog will payout slower but ultimatley (unless pool collapses) will payout the same amount as a pool with no backlog. Traditionally hopping has been seen as a method to get an unfair reward (reward greater than PPS value w/ no fees). Hopping SMPPS pools doesn't achieve that goal. – DeathAndTaxes – 2011-10-17T16:17:16.273

The "traditional" definition is too limited. Several factors affect attractiveness of a pool, expected value being the foremost but variance and maturity time are important too. "Hopping" should refer to any situation when a pool has different attractiveness at different times, and people take advantage of this by mining only at attractive times, getting an unfair advantage over continuous miners. Unit-PPLNS, double geometric and so on are hopping-proof - the attractiveness is exactly the same at all times. – Meni Rosenfeld – 2011-10-17T18:47:13.247

SMPPS and ESMPPS (Equalised SMPPS) are "hoppable" only in the sense that they don't explicitly punish hoppers. They are resilient against the prevalent style of hopping but they do nothing to lessen the gains of users who hop away. For this reason, SMPPS (and ESMPPS) are considered "fair" algorithms by hoppers and pools using E/SMPPS are often chosen as "backup pools" to be mined at when no other pools meet their specific criteria. Algorithms like PPLNS (Pay Per Last N Shares) punish anyone who isn't a 24/7 miner regardless of when or how they choose to hop in/out and so aren't even viable as backup pools.

In short, it's not realistic (to my knowledge) to say that SMPPS could be in any way gamed by pool hoppers. They are, however, fair to hoppers who choose to use them as a backup pool and so many who are staunchly anti-hopping still consider SMPPS to be bad/inadequate.

David Perry

Posted 2011-10-13T18:21:43.140

Reputation: 13 848

+1 for solid answer and clarification. Using terms like "hoppable" are ambiguous. Proportional (and some score based) pools benefit pool hoppers. SMPPS (and variants) are neutral with hoppers no better or worse than 24/7 miners. PPLNS (and variants) hurt pool hoppers – DeathAndTaxes – 2011-10-14T19:27:00.970

7 years later and I've randomly encountered this question. I now have enough reputation to comment on this answer. So for posterity, I'd like to point out that this answer is very wrong, as I've explained in a comment on my own answer - https://bitcoin.stackexchange.com/questions/1508/can-an-smpps-shared-maximum-pay-per-share-pool-be-hopped#comment1986_1564.

– Meni Rosenfeld – 2018-10-25T00:54:20.467