Pool Reward Comparison

[GimEEE]

Obviously one lone miner getting gypped can't create a streissand effect, but I'm merely attempting to create a thread for discussion of those who have experience with different miner reward systems.

First, I'd like to thank artifact2 for being so gracious in hosting these really awesome statistics on the eligius pools --> +1

1) The standard proportional payout is Contributed shares % times 50 BTC (or block shares given / block shares total x 50 btc)
so, if you contribute 1% of a certain block round's shares, you get rewarded .5BTC, .5% = .25BTC, etc. http://www.btcguild.com/index.php

2) Since the eligius pps hit me in the wallet around June 14th, I've noticed Eligius PPS consistently rewards less than the proportional expectation. For a sample address given on the stats page http://eligius.st/~artefact2/blocks/16U5HHxjTxp4ir1NhDr3F4jnqGCScRJwvt, we can calculate this simply - eligius pps reward over first 4 rounds of "s3" = 0.03632554 + .02131669  + .10553334  + .12322369  =  ~.286 BTC
But, the reward if this was proportional would be (0.2382 + .0399 + .2578 + .2464) / 100% x 50 btc = .391 btc expected

So, a difference of .391 - .286 ~ .105 BTC (out of .391 expected) for a proportional loss of  ~ 27%

Hopefully this loss will approach zero over time, or whatever cut the pool operators coded for themselves.
***So, if you plan to mine forever under a wtfpps system, you might not have a significant reduction in payout, but you will NEVER have a higher payout than proportional in short or long term scenarios.

But either way, it's good to get these facts out in the open for potential miners before they choose where to submit their work.
I'd appreciate it if responders stick to the facts, rather than emotional attachment to the pool or it's operators.

*edit

[1715025692]

1715025692

[1715025692]

1715025692

[1715025692]

1715025692

[1715025692]

1715025692

[1715025692]

1715025692

[gentakin]

I've noticed Eligius PPS consistently rewards less than the proportional expectation.

It is easy to show that your claim is false.

Check the stats you linked. See block 0000000000000BE284219936487F7A2310CD7D795FF7EA5E40C57AA1AE18C85C?
It says "Contribution 0.0399 %".
0.0399% of 50BTC = 0,01995BTC. However, the guy earned 0.02131669BTC. So he got more than the proportional expectation.

If you keep watching the stats, you will notice the same for any round longer than average (i.e. more than 1,379,223 shares for now), unless the pool happens to be on a longer period of bad luck.


Please read (and understand) about MaxPPS before making these claims. But it's nice to see you want to stick to the facts from now on. (I do understand you're upset because the eligius US payouts took pretty long. However your claims about MaxPPS are wrong.)

[Luke-Jr]

1) The standard proportional payout is Contributed shares % times 50 BTC (or block shares given / block shares total x 50 btc)
so, if you contribute 1% of a certain block round's shares, you get rewarded .5BTC, .5% = .25BTC, etc.

Your first flaw is in assuming there's something inherently fair about the Proportional method. This usually comes from the misconception that people can make progress mining. The fact is that over 99.99% of shares are completely worthless and do absolutely nothing for the pool. They have no relevance whatsoever to the block that might (or might not) be found at the end of the round. Only 1 share finds that block, by itself, with no help from the other shares.

The goal of pools is to pay miners for lower-difficulty fake "blocks", while collecting the rewards itself from the real blocks. This reduces variation for the miners. Proportional is simple, but by far not the only way pools might decide to pay miners. The most fair method, with least variation, is to pay them 1/difficulty for each difficulty 1 share: this is called Pay Per Share. In comparison, Proportional pays miners more than their fair (PPS) reward when it finds a block quickly, and less when it takes a long time to find a block. Because of this uneven distribution, it is most effective to implement "pool hopping" to mine on the pool currently offering the highest reward per share (ie, the one with the fewest shares in the current round).

Straight PPS was originally implemented by OneFixt2 for BitPenny, and is a popular option at Deepbit despite their 10% fees. However, it is more vulnerable to true cheating, and when this happens, the pool operator takes the infinite hit from it.

Eligius has (after much notice and even poll from members) switched to a system almost like PPS. Miners are generally paid the full 1/difficulty reward for each share (without the huge 10% fees historically associated with PPS), but the system keeps track of what the pool has itself earned overall, and limits payouts to ensure the operator does not take a hit when there is cheating or extreme unluckiness going on. At the same time, it tracks what miners would have earned under straight PPS, and if someone was ever underpaid (due to not having the funds available) will pay them a bonus later if it has extra funds (from lucky/short blocks).

All the payout systems will eventually tend toward the same total payout over time, with one exception: due to the proportional method's predictable unfairness pattern, miners can achieve 30% greater earnings than their work's fair pay. Those who do not implement pool hopping, see an effective 30% loss over time. This is the primary motivation for avoiding the proportional system.

[Yeti]

Thanks, Luke-jr for clearing that up!

For the first two blocks I though "oh, no advantage from lucky blocks anymore... how sad". But now the current block has already over 2,000,000 shares and my estimated payout is still rising! Usually I only get 0.007 BTC from each block but now it shows 0.012 BTC.
That's the fairest version I can think of, which is probably why so many people voted for it.

If you want to screw over the pool, use one with Straight PPS! If you want to participate in lucky blocks, solo mine! If you want nice constant payouts from your constant Mh/s, come to Eligius with MaxPPS!!!

Edit: Aaaand I just got my first payout from Server 3 in block 133729, 0.34 BTC! Thanks a lot!

[twmz]

Notice that in the most recent round, my contribution was just below 1%.  In proportional (btcguild, deepbit prop, etc), I would have gotten only 0.49885 BTC.  I actually got 0.77635355 BTC.

The system works.

[GimEEE]

Notice that in the most recent round, my contribution was just below 1%.  In proportional (btcguild, deepbit prop, etc), I would have gotten only 0.49885 BTC.  I actually got 0.77635355 BTC.
The system works.

I think I see, the sacrifice of higher reward in previous rounds (compared to proportional), is compensated by a smooth increase in rewards on later long blocks. Kinda like paying it forward to the pool. I like it!

Pool hopping being profitable is not all too feasible afaik. Has anyone seen any hopper scripts?
The profitable scenario would require the hopper to
a) constantly keep his miners at high speed (to compete with steady miners going full blast)
while
b) constantly mining on the freshest proportional pool (to get the pool-hop effect)
and also minimize lost shares from hopping.
This seems a little hard to code, and possibly harder to effectively implement without designing your own miner from ground up.
Wouldn't the hopper lose many shares from having the miner constantly switch pools every time it identified a "fresh pool", they lag from having to switch gears and reallocate it's efforts.

Not to mention their rewards are split between so many pools with various minimum payouts, etc. would be a pain to manage.

Theoretically, it is possible, I guess, but further isn't pool hopping easily detectable by the targeted proportional pool?
They'll see the miner has consistently lower rewards shares on all longer blocks, which I'm sure they'd notice if this pool hopping script existed.
Having open stats like artifact2 is definitely a good way to catch it though with more eyes on miners in general.

Also, couldn't the pools easily code the server to give shares to the oldest miners first (LP, etc.), thus rewarding the steady miners who remain loyal, while also allowing the hoppers to jump in and contribute but last in the queue to get work.
You talk about a 30% loss, but also as you say, every share of work done is a possibility for a solve, it's just as possible that letting the hopper jump in could result in a block solve from that extra work, thus actually benefit the pool.

Seems a pool could easily punish a hopper without punishing the whole pool, such as one pool that cuts 50% of the reward for miners who aren't working in any of the last 30 minutes of the block solve.
Basically, if hopping were a problem, there seems simple technical solutions.

[gentakin]

Pool hopping being profitable is not all too feasible afaik. Has anyone seen any hopper scripts?

Yes. Check it out: http://forum.bitcoin.org/index.php?topic=17970.0

This seems a little hard to code, and possibly harder to effectively implement without designing your own miner from ground up.

It's not really hard. poclbm and phoenix can be modified to hop quite easily if you know what you're doing.

Wouldn't the hopper lose many shares from having the miner constantly switch pools every time it identified a "fresh pool", they lag from having to switch gears and reallocate it's efforts.

A getwork request (which is everything needed to switch pools) takes approx. 1 second. (edit: Clearing the result/getwork queues and forcing the miner thread to stop work should be easy to do as well, so you don't submit "old shares" to the new pool.)

Not to mention their rewards are split between so many pools with various minimum payouts, etc. would be a pain to manage.

Most proportional pools don't have a minimum payout. Only BitClockers comes to my mind, they have 0.10 (that would be a lot for me, especially with pool hopping, but I'm only at 280mhash/s).

Theoretically, it is possible, I guess, but further isn't pool hopping easily detectable by the targeted proportional pool?

That's right - but they can't prove that you're hopping. You might just be lucky and have connection problems on long rounds. (Okay, sounds like a bad excuse   )
Also, I suspect most pool operators wouldn't like banning hoppers. After all, hoppers might solve a block as well, giving the operator 0-3% fees and transaction fees. Only the other miners are getting less. (They don't mind obviously, as proportional is still the favourite reward scheme.)

Also, couldn't the pools easily code the server to give shares to the oldest miners first (LP, etc.), thus rewarding the steady miners who remain loyal, while also allowing the hoppers to jump in and contribute but last in the queue to get work.

Hopping would still be profitable with this, I guess, as most pools are not close to being overloaded.

You talk about a 30% loss, but also as you say, every share of work done is a possibility for a solve, it's just as possible that letting the hopper jump in could result in a block solve from that extra work, thus actually benefit the pool.

It's probably not 30% yet, as only few miners are skilled enough/care enough to pool hop.
And it benefits the pool, yes, but the pool benefits the hopper more. Hopping is a win for the hopper, and it's a win for operators of small pools that get some GH/s boost in the beginning of rounds. It's a loss for non-hopping miners.

Seems a pool could easily punish a hopper without punishing the whole pool, such as one pool that cuts 50% of the reward for miners who aren't working in any of the last 30 minutes of the block solve.

If I understand this correctly, it should be easy to send a few shares to the pool every 30min, thus keeping them from cutting your reward in half. Also, I seem to remember that this certain pool allows to get rid of the 50% cut if you donate a certain percentage of your reward (like 2% or 3%) - a pool hopper easily gets a bouns of more than 2% or 3%.

[GimEEE]

thanks, that definitely clears up the theory of hopping.
Maybe a simple 5 minute delay for the 2 pools publishing the shares submitted per round would effectively kill the hoppers profitability, eliminating the hopper benefit for the first 5 min.
Also, maybe the pool could throttle new miners for the first 5-10 minutes of each round. With btcguild/deepbit, this is almost half the round anyways.
Or, a miner entering at the beginning of a round only gets work as fast as they were at the end of the last round for the first 5-10 minutes. This would actually decrease the hoppers work and result in them not submitting shares effectively.

But, as I said, monitoring stats would be the best way to catch the hopper, rather than sacrifice non-hoppers having payouts far below the expected proportional payout.
Like this guy: http://eligius.st/~artefact2/3/16ccjkuuQjQ64H9qssmXnj695DdBDR75wJ
WTF, submitting 20% of block shares, but with a  0.4 BTC reward.
Must be the pool op (who gets the lost rewards anyway), or is there another rational explanation for why they keep mining there? (yeah botnet that can't be redirected to better paying pool, maybe. . .)
But that miner did get rewarded 0.73 in that block generation: http://blockexplorer.com/block/00000000000002ff87f817e00d3f86e063c9caa16e7c99ce329b7f3347e4ecf7
with 48.70949313BTC being held by the pool operator. . .
more fair than proportional payouts?

[Luke-Jr]

WTF, submitting 20% of block shares, but with a  0.4 BTC reward.

He submitted 0.4 BTC worth of shares, so he gets paid 0.4 BTC. Simple. Your percent is an arbitrary number, with no relevance to the work done.

[gentakin]

Eligius is one of the best-paying pools out there for non-hopping miners (assuming there are no idle shares, downtimes or other technical problems).

Stop comparing single round rewards to their proportional equivalent. You should compare reward per time span (i.e. reward for 24 hours mining). Then, eligius will turn out to pay more than i.e. Deepbit (because of 3% fee and maybe some pool hoppers who exploit deepbit. Though I wouldn't know how to do it, as deepbit delays stats by 1 hour.). Depending on the number of pool hoppers at BTC guild, eligius will also pay more than BTC guild, even at 0% donation.

Of course, if you start 2 miners with equal hashrate right now and look at them tomorrow, you might earn more at deepbit if it gets lucky (check their stats page). If deepbit has average luck, and eligius is not currently on a longer period of bad luck (where it can't pay from savings), then you will get more BTC from eligius than deepbit.


(BTW - right now, eligius appears to be *very lucky*, with only very few blocks >1,000,000 shares. So the savings are probably big enough to suffer through a period of bad luck with longer rounds.)

[klaaster]

As you posted to my chart in the other thread here is an update.
The system performs exactly es expected, I really like it.

Thank you Luke-Jr and Artefact2 for your work, smart Pool and great stats.  

[anodyne]

Th 60Gh/s second user is to consistent to be a botnet, the idle time of a research cluster or even a smaller pool that runs off Eligius, so it's probably a corporate entity who uses the pool either for its own purpose or (maybe more likely) for rented mining contracts. There was a miner with a similar hashrate on the EU pool, but who dropped off at the beginning of Eligius-3... so, if we assume that it is the same user who has returned I guess we can say that someone with very large investments in mining hardware and a sense of business has given the system his approval after watching it for a day or so.

[GimEEE]

Th 60Gh/s second user is to consistent to be a botnet, the idle time of a research cluster or even a smaller pool that runs off Eligius, so it's probably a corporate entity who uses the pool either for its own purpose or (maybe more likely) for rented mining contracts. There was a miner with a similar hashrate on the EU pool, but who dropped off at the beginning of Eligius-3... so, if we assume that it is the same user who has returned I guess we can say that someone with very large investments in mining hardware and a sense of business has given the system his approval after watching it for a day or so.

Ouch, almost painful to look at - http://eligius.st/~artefact2/3/16ccjkuuQjQ64H9qssmXnj695DdBDR75wJ
But it doesn't smooth the reward down to much less than the expected for his hash rate, I'll concede that.
It'll be interesting to see if the miners get caught up on payments with the long blocks coming without having to wait a week or two.


This new reward system definitely clears up the problem of pool operator not getting paid enough (quite a windfall - 34+ out of 50 btc on this one - http://blockexplorer.com/block/00000000000003bc639f3a98f337d105706583c0dc383ef75bc508240c6ae91c)

Seems the key to eligius pool operators windfall is going to be if potential miners don't get access to the facts of the payout system we've discussed in this thread. i.e. moving this thread from mining/pools to a more concealed forum location will likely help you hit more noobs with your pps system. Willing to use any tricks you can to try to make some BTC I see :-P

[anodyne]

You're still not getting the system... there is no smoothing on short rounds. It's pay per share without a fee. The smoothing happens when there's a round that's so long that there are no accumulated funds to pay full PPS.

PPS value is based on the expected amount of shares needed for a block, but since that involves a lot of decimals we'll say that the expected number of shares is 100 per block, so they are each worth 0.5BTC. So, it would go something like this:

Block 1: Solved after 50 shares. Pool pays out a total of 25BTC
Block 2: Solved after 150 shares. Still fine, since the pool has 25BTC left and can still pay the total 75BTC
Block 3: Solved after 50 shares. Pool keeps 25BTC again.
Block 4: Solved after 200 shares. Not enough to pay 0.5 for each share, so the 75BTC are distributed according to amount of shares contributed, but weighted by work contributed to earlier blocks.

So, it fixes the the problem of normal PPS where the pool takes  an 8-10% fee on each share to cover for times of bad luck , and where the pool makes a profit if there are no such blocks.

Edit: Or, you could go look at the round with almost 4M shares, where Mr. 60Ghash got paid 27.8BTC for a 19% contribution, and with some multiplication of that percentage I would assume that the pool paid for a total of about 140BTC worth of shares in that round.

[GimEEE]

Edit: Or, you could go look at the round with almost 4M shares, where Mr. 60Ghash got paid 27.8BTC for a 19% contribution, and with some multiplication of that percentage I would assume that the pool paid for a total of about 140BTC worth of shares in that round.

The page I'm looking at shows the pool operator has withheld payment of over 28BTC of "earned" reward to this business savvy miner.
http://eligius.st/~artefact2/3/16ccjkuuQjQ64H9qssmXnj695DdBDR75wJ
I hope Eligius does pay this hard working miner everything they owe (if this isn't the op's miner anyways).
I did get paid by eligius eventually, but it took about 2 weeks, then the pool op rated me negative for publicizing his practices  :-(
Not to mention the forum mod diablo (and possibly others) abusing his authority to censor this thread and threads that link to it :-(
http://forum.bitcoin.org/index.php?topic=23760.0

[Yeti]

You still don't get it, do you? Eeeverybody else is wrong and you're the lonely cowboy fighting for an earnest man's pay.

Stop mining at Eligius this instant and/or stop complaining or I will sue you for mental cruelty!!

[GimEEE]

You still don't get it, do you? Eeeverybody else is wrong and you're the lonely cowboy fighting for an earnest man's pay.
Stop mining at Eligius this instant and/or stop complaining or I will sue you for mental cruelty!!

LOLS, where was I wrong?
Why was this thread censored?

[Yeti]

It wasn't. It's a glitch in the forum software. Now please move along, there's nothing to see here.

[anodyne]

I'm pretty sure they will be paid out, once the blocks are confirmed and everything.

I can see how the situation with the US pool being closed in the middle of testing a new system could be a bit upsetting, but I can also agree with those that think that you're causing a bit too much noise over this. For example, I'm pretty sure the Eligius thread still had "please test... experimental pool" in the title when this was going on, and I think that would indicate that things should be expected to not work properly all the time.

[Yeti]

Plus the fact that GimEEE expects Priority Support for Professional Miners (sm) when he doesn't actually pay anything for it and this pool is little more than a fancy hobby. He should just go to one of these pools that are a business and get a contract with penalties in it for "skimming miners" and "withholding fair compensation" for "honest hard working ordinary miner people", yadda yadda yawnda.