[OFFLINE] SIMPLECOIN.US [PPLNS/SMPPS]

[GoldenBoar]

RyNinDaCleM was spot on with his analysis.

Say I decide I want to do some gaming, and I stop mining before N, I don't get any reward at all. PPLNS only makes sense for 24/7 miners. It's also vulnerable to reverse pool hopping. Say you and the pool hooper had the same hashrate, you could mine the whole round and only get the same reward as the pool hooper who mined the last half (assuming N = round shares/2). You do twice the work for the same reward.

What do you mean by "and I stop mining before N"?

If you decide to stop for gaming then your shares don't disappear.  Each share will generate funds for you for any and all blocks which are found by the pool during the pool's next N shares (whether or not you mine for this period is irrelevant).  If the pool finds no blocks during the N shares following your submission then your share has failed to generate any BTC for you at all (again, whether or not you mined at the pool during this time is irrelevant).

PPLNS means that when a block is found, the reward is split uniformly between the most recently submitted N shares.  Such a reward system neither rewards nor punishes intermittent miners in terms of average income.

In PPLNS, N is normally either constant or tied to difficulty, not to "round shares".  If N were "round shares" then the reward system would suffer the usual pool hopping problem faced by proportional miners.  When N is 1 the reward system is identical to solo mining (which also neither punishes nor rewards intermittent miners in terms of average income).

Absolutely, this reward system is vulnerable to "reverse pool hopping".  Shares at the end of a round are more valuable than shares at the beginning in general.  If a hopper hops into the pool for the last N shares of each round then they are going to earn much more than a 24/7 miner or random intermittent miner.  However, this strategy requires the ability to predict the future.

From Eligius:

A pool accepts shares constantly. It assigns each share a number based on its order of arrival. It also decides on a number N, the 'look back constant'. Whenever a valid block is found at share X, the contributors of all prior shares in the range (X-N) exclusive to X inclusive are credited 50/N BTC.

So, if I stop mining and none of my shares end up within the range (X-N) then I get no reward.

I know N isn't tied to round shares, that was just for the example I used. Lets say N = difficulty/2. In order to reverse pool hop, all I need to do is find a pool whose round shares are greater than N. Basically the more shares past N the better.

[1713441732]

1713441732

[1713441732]

1713441732

[1713441732]

1713441732

[teukon]

From Eligius:

A pool accepts shares constantly. It assigns each share a number based on its order of arrival. It also decides on a number N, the 'look back constant'. Whenever a valid block is found at share X, the contributors of all prior shares in the range (X-N) exclusive to X inclusive are credited 50/N BTC.

So, if I stop mining and none of my shares end up within the range (X-N) then I get no reward.

I know N isn't tied to round shares, that was just for the example I used. Lets say N = difficulty/2. In order to reverse pool hop, all I need to do is find a pool whose round shares are greater than N. Basically the more shares past N the better.

The description you quoted from Eligius is spot on; this is the basic idea behind PPLNS.  A real implementation will be slightly more complicated because it must account for changes in N and/or in difficulty and must also be careful about what happens when the pool starts and ends.

Yes, if you stop mining and none of your shares end up within the (X-N, N] range then you get no reward.  If there is one block in the range (X-N, N] you get 50/N BTC (assuming no transaction fee rewards and 0% fee), if there are two blocks in the range (X-N, N] you get 100/N BTC, and so on.

Let's suppose we have a PPLNS pool with N = difficulty/2 (and assume difficulty is fixed for simplicity).  The PPLNS pool here is a good real world example.  The average number of shares to a round here is 2*N (because share difficulty is 1 here).  Some rounds are much shorter and others are much longer but most are longer than N.  It is not possible to hop this pool in the way you describe without some kind of time machine.  To "reverse pool hop" you need to know when a round is going to end.

[RyNinDaCleM]

Whew! I didn't mean to start a heated argument! XD

Yes! On average, the payout would be the same, no matter where the shares go. I do believe however, part time mining at PPLNS is a bigger gamble. If you work during the first N shares of a round, and decide to stop to play a game, from that point on your shares start losing value. If during this time a big miner comes in, and in the next few hours another N shares are submitted, your shares would be a complete waste.

It is not possible to hop this pool in the way you describe without some kind of time machine.  To "reverse pool hop" you need to know when a round is going to end.

That falls in the same kind of probability as hopping a prop pool. If you know when N comes and goes in a specific round, you have better odds of not *over working* for the same payout. Now, it may not work every time, but just like standard hopping, it probably works enough to make it worth the trouble.

[teukon]

Whew! I didn't mean to start a heated argument! XD

Lol, sorry about that.

Yes! On average, the payout would be the same, no matter where the shares go. I do believe however, part time mining at PPLNS is a bigger gamble.  If you work during the first N shares of a round, and decide to stop to play a game, from that point on your shares start losing value. If during this time a big miner comes in, and in the next few hours another N shares are submitted, your shares would be a complete waste.

The value of your shares changes as more information about the near future of the pool is revealed.  This has nothing to do with whether or not you continue to mine at the pool yourself.  If I submit a share it will lose value as non-block-generating shares are submitted and gain value when blocks are found; N pool shares after my submission the true value of my share will be fully revealed.  Exactly the same happens at proportional pools and, indeed, with most reward systems.  A share is behaving very much like the scratch card I described which has a 1/4 chance of being a winner ($4 prize).  The value of the scratch card is $1 (expected value) but, when it is scratched to reveal no prize its value decreases $0.  The risk of this decrease in value remains whether or not you decide to buy more cards before scratching the first.

The only two sound ways of mining I'm aware of which don't decrease the value of your share as future non-block-generating shares are submitted (i.e. have zero share variance) are PPS and PPLNS with N = 1 (effectively solo mining).  These are both completely hop-proof and don't favour 24/7 mining over intermittent mining.  PPS requires a fee to be sustainable (at 0% the pool balance will follow a random walk and will become arbitrarily large and negative at points in the future).  This problem cannot be avoided (SMPPS tries to avoid it by placing the random walk along the time axis - this is more subtle but equally flawed).

It is not possible to hop this pool in the way you describe without some kind of time machine.  To "reverse pool hop" you need to know when a round is going to end.

That falls in the same kind of probability as hopping a prop pool. If you know when N comes and goes in a specific round, you have better odds of not *over working* for the same payout. Now, it may not work every time, but just like standard hopping, it probably works enough to make it worth the trouble.

I agree that the mechanic for hopping is similar to that for a proportional pool.  However, a proportional pool can be effectively hopped with information from the past whereas, a PPLNS pool can only be exploited using information from the future.  I agree that you don't need to know exactly when the round will end but the lack-of-memory property of the distribution of future blocks w.r.t. shares tells us that the current round duration gives us absolutely no information on the expected number of blocks in the next N shares.

[GoldenBoar]

Yes, if you stop mining and none of your shares end up within the (X-N, N] range then you get no reward.  If there is one block in the range (X-N, N] you get 50/N BTC (assuming no transaction fee rewards and 0% fee), if there are two blocks in the range (X-N, N] you get 100/N BTC, and so on.

Surely you can see then, why I would never use such a system?

Let's suppose we have a PPLNS pool with N = difficulty/2 (and assume difficulty is fixed for simplicity).  The PPLNS pool here is a good real world example.  The average number of shares to a round here is 2*N (because share difficulty is 1 here).  Some rounds are much shorter and others are much longer but most are longer than N.  It is not possible to hop this pool in the way you describe without some kind of time machine.  To "reverse pool hop" you need to know when a round is going to end.

If you look at the PPLNS stats page, you can see that the number of BTC shares this round is 2,140,994 (at the time of posting) and the difficulty is 1,468,195.430. If I start mining this pool now, then it's likely that all my shares will be within (X-N), therefore, I'll get paid a lot more for the work I put in than someone who has been mining since the beginning of the round. Reverse pool hopping actually seems pretty easy to me.

[teukon]

Yes, if you stop mining and none of your shares end up within the (X-N, N] range then you get no reward.  If there is one block in the range (X-N, N] you get 50/N BTC (assuming no transaction fee rewards and 0% fee), if there are two blocks in the range (X-N, N] you get 100/N BTC, and so on.

Surely you can see then, why I would never use such a system?

No I don't.  What is wrong with it?

If you look at the PPLNS stats page, you can see that the number of BTC shares this round is 2,140,994 (at the time of posting) and the difficulty is 1,468,195.430. If I start mining this pool now, then it's likely that all my shares will be within (X-N), therefore, I'll get paid a lot more for the work I put in than someone who has been mining since the beginning of the round. Reverse pool hopping actually seems pretty easy to me.

If you start mining now then your chances of your shares being in the range (X-N, N] are exactly the same as they would be at the beginning of a round.  The expected round length for the current round is precisely 2,140,994 + 1,468,195.430 = 3,609,189.43 shares (basic corollary of the lack-of-memory property) and it gets longer with every single non-block-generating share submitted.  Suggesting that the round is close to an end because many shares have been submitted is like suggesting that a heads is due after flipping a fair coin five times and getting tails each time.

You are more than welcome to try to "reverse pool hop" the PPLNS pool here, you will be helping the miners out by lowering the variance slightly and you'll get a fair reward for your shares in the long run.

[GoldenBoar]

Yes, if you stop mining and none of your shares end up within the (X-N, N] range then you get no reward.  If there is one block in the range (X-N, N] you get 50/N BTC (assuming no transaction fee rewards and 0% fee), if there are two blocks in the range (X-N, N] you get 100/N BTC, and so on.

Surely you can see then, why I would never use such a system?

No I don't.  What is wrong with it?

Precisely what you said:

Yes, if you stop mining and none of your shares end up within the (X-N, N] range then you get no reward.

If you look at the PPLNS stats page, you can see that the number of BTC shares this round is 2,140,994 (at the time of posting) and the difficulty is 1,468,195.430. If I start mining this pool now, then it's likely that all my shares will be within (X-N), therefore, I'll get paid a lot more for the work I put in than someone who has been mining since the beginning of the round. Reverse pool hopping actually seems pretty easy to me.

If you start mining now then your chances of your shares being in the range (X-N, N] are exactly the same as they would be at the beginning of a round.  The expected round length for the current round is precisely 2,140,994 + 1,468,195.430 = 3,609,189.43 shares (basic corollary of the lack-of-memory property) and it gets longer with every single non-block-generating share submitted.  Suggesting that the round is close to an end because many shares have been submitted is like suggesting that a heads is due after flipping a fair coin five times and getting tails each time.

Do the math when the round then ends then. We would get paid the same amount but I would have done less work.

[teukon]

Precisely what you said:

Yes, if you stop mining and none of your shares end up within the (X-N, N] range then you get no reward.

This is not a problem, it is a fact about the reward system.

Do the math when the round then ends then. We would get paid the same amount but I would have done less work.

This is true but completely irrelevant to the issue of pool hopping.

GoldenBoar, your recent comments betray your lack of understanding and it's clear to me that you won't comprehend the mathematics without some careful thought and study.  If you genuinely want to learn then feel free to PM me.

I'm sorry Mike for hijacking your thread, I got carried away.  I'll stop now.

[GoldenBoar]

Precisely what you said:

Yes, if you stop mining and none of your shares end up within the (X-N, N] range then you get no reward.

This is not a problem, it is a fact about the reward system.

How is not getting rewarded for my efforts not a problem? Would you aslo consider it not a problem if I hired you to do a job and then refused to pay you for the work? To me, it's a major problem and that's why I'll never use PPLNS.

Do the math when the round then ends then. We would get paid the same amount but I would have done less work.

This is true but completely irrelevant to the issue of pool hopping.

How can that possibly be irrelevant? That's the whole point of pool hopping - to gain a higher reward than usual for the amount of work put in.  

[simplecoin]

Wow, I've missed a lot while I was ill.

That said, I've monitored the site from my tablet, and it's been holding stable (although the hashrate is all over the place).

Yes, in theory you can game PPLNS, however it wouldn't make sense with proportional pools available. Additionally, the short rounds of pplns would all be missed by pplns hopping, making it exceptionally unattractive to pool hoppers.

[teukon]

Wow, I've missed a lot while I was ill.

That said, I've monitored the site from my tablet, and it's been holding stable (although the hashrate is all over the place).

Yes, in theory you can game PPLNS, however it wouldn't make sense with proportional pools available. Additionally, the short rounds of pplns would all be missed by pplns hopping, making it exceptionally unattractive to pool hoppers.

Sorry to hear you were ill.  I'm glad to see that the pool is behaving properly now (blocks are appearing just as with the other merged mining pools).

It's regrettable that the hashrate is down but that at least means the miners here will be enjoying the promotions for longer.

[GoldenBoar]

There's something wrong with the SMPPS pool. The current BTC round share is ~100,000 but no block is being listed as found. If we haven't found a block, it should be ~3,000,000.

[simplecoin]

There's something wrong with the SMPPS pool. The current BTC round share is ~100,000 but no block is being listed as found. If we haven't found a block, it should be ~3,000,000.

quite right, I'm looking into it. No block in the transaction log.

[GoldenBoar]

There's something wrong with the SMPPS pool. The current BTC round share is ~100,000 but no block is being listed as found. If we haven't found a block, it should be ~3,000,000.

quite right, I'm looking into it. No block in the transaction log.

Everything else seems to working right. Some of my rewards have been shifted from Unconfimed to Balance.

[hmblm1245]

pool's been stable since the hack.Good work. Thanks.

[simplecoin]

Seems like there is an issue in block reporting since I rebuilt the server. All winning blocks are being rewarded properly, but not all are being diplayed.

There WAS an SMPPS BTC block found and paid, but it is not showing.

I've made a change to the database to facilitate a fix, but it may take a few days to straighten out the old blocks.

UPDATE: Or NOT I managed to implement the new block dates very quickly. There shouldn't be any missed by stats now and the page load on stats and blocks is MUCH faster.

Now I can maybe focus on my failing headless miner.

[RyNinDaCleM]

Lol! I didn't know computers still did the whole 1900's thing anymore. What, no BTC flying out of the BitCoin ATM?

[hmblm1245]

Lol! I didn't know computers still did the whole 1900's thing anymore. What, no BTC flying out of the BitCoin ATM?

Y2K!!!!

[simplecoin]

Lol! I didn't know computers still did the whole 1900's thing anymore. What, no BTC flying out of the BitCoin ATM?

Y2K!!!!

lol, yeah...... php time() doesn't like being dropped into strings..... it's fixed for future blocks.

That said, LTC support is incoming. Any preference to pool type? PPLNS or Prop?

[hmblm1245]

I would say PROP for LTC. So many crypto currencies i can't keep up.