anth0ny
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 12:47:33 AM |
|
A big question that needs to be answers first is what do you do if the log goes empty.
This can already happen, in theory, if the pool gets really lucky. What happens when this occurs is not described, as far as I can tell. Crediting shares at effectively 100% PPS even though the actual earnings must be less than that reduces the probability of the share log running empty, but that strikes me as somewhat sweeping the problem under the rug. If the entire share log is paid the excess goes to a buffer. Then when the next block is found, the amount of the buffer is added to its reward for purposes of paying the share log. This is described here.Well you agree that people have a greater expected value if they mine while there's a big buffer than if they mine when there's no buffer, right? Imagine there's a 2000 BTC buffer when you mine your shares. Your expected value is going to be pretty damn close to 100% under the current system, right? Not a measly 98%.
|
|
|
|
roy7
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 01:03:56 AM |
|
If there were lots of CPPSRB pools then you could engage in hopping to the ones with the smallest backlogs I guess. ![Wink](https://bitcointalk.org/Smileys/default/wink.gif)
|
|
|
|
wizkid057 (OP)
Legendary
Offline
Activity: 1223
Merit: 1006
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 01:05:54 AM |
|
PPLNS variance is higher than CPPSRB. <snip>
Is this from simulation or analysis? Merni Rosenfeld derives the reward variance for PPLNS on page ten here: https://bitcoil.co.il/pool_analysis.pdfFor PPLNS, the variance is 1/(network mining difficulty)*(Bitcoin block reward)^2 * 1/N (N is the "N' in PPLNS) What is the variance you've calculated for CPPSRB (analytical or otherwise)? Did you take into account the varying maturity time for CPPSRB compared with the more predictable maturity time for PPLNS? Cheers, wk - I've been interested to know this for ages. I ran simulations of multiple reward systems against real-world data (Eligius shares database/blocks) prior to implementing CPPSRB. (I've rerun this a few times since then, but not recently.) While I admittedly do not know how to boil it down to a single equation, I'm pretty good at simulation ![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif) I took short and long term averages and % deviations from maximum/expected PPS for various time frames accounting for individuals as well as pool-wide earnings with multiple reward systems. While PPLNS (with various different N's) ended up pretty close to CPPSRB's % deviation quickly, CPPSRB always tended to win by at least a few %. Over the longer term and taking into account the real world data (where some miners were not consistent for example) PPLNS left some miners with substantially higher variance than others, while CPPSRB managed to iron this out quite well in most cases. In my simulations CPPSRB kept more miners closer to 100% PPS than PPLNS did over almost all time frames, short and long. -wk
|
|
|
|
anth0ny
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 01:06:20 AM |
|
If there were lots of CPPSRB pools then you could engage in hopping to the ones with the smallest backlogs I guess. ![Wink](https://bitcointalk.org/Smileys/default/wink.gif) Or just hop to PPLNS whenever the backlog is less than zero. ![Wink](https://bitcointalk.org/Smileys/default/wink.gif) (In reality there are other factors which are probably more important, though. Not the least of which is fees.)
|
|
|
|
organofcorti
Donator
Legendary
Offline
Activity: 2058
Merit: 1007
Poor impulse control.
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 01:10:26 AM |
|
I ran simulations of multiple reward systems against real-world data (Eligius shares database/blocks) prior to implementing CPPSRB. (I've rerun this a few times since then, but not recently.) While I admittedly do not know how to boil it down to a single equation, I'm pretty good at simulation ![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif) I took short and long term averages and % deviations from maximum/expected PPS for various time frames accounting for individuals as well as pool-wide earnings with multiple reward systems. While PPLNS (with various different N's) ended up pretty close to CPPSRB's % deviation quickly, CPPSRB always tended to win by at least a few %. Over the longer term and taking into account the real world data (where some miners were not consistent for example) PPLNS left some miners with substantially higher variance than others, while CPPSRB managed to iron this out quite well in most cases. In my simulations CPPSRB kept more miners closer to 100% PPS than PPLNS did over almost all time frames, short and long. -wk Thanks for putting in the work, wk. I reckon you should put your data on Eligius' website, maybe publish it as an experiment to make it more replicable? Or maybe you have a pool to run ![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif)
|
|
|
|
organofcorti
Donator
Legendary
Offline
Activity: 2058
Merit: 1007
Poor impulse control.
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 01:13:07 AM |
|
If there were lots of CPPSRB pools then you could engage in hopping to the ones with the smallest backlogs I guess. ![Wink](https://bitcointalk.org/Smileys/default/wink.gif) I'd be interested to see how much you could shorten the expected time to maturity using this approach for an arbitrary number of CPPSRB pools.
|
|
|
|
roy7
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 01:21:17 AM |
|
If there were lots of CPPSRB pools then you could engage in hopping to the ones with the smallest backlogs I guess. ![Wink](https://bitcointalk.org/Smileys/default/wink.gif) I'd be interested to see how much you could shorten the expected time to maturity using this approach for an arbitrary number of CPPSRB pools. Let us know how it turns out. ![Wink](https://bitcointalk.org/Smileys/default/wink.gif)
|
|
|
|
anth0ny
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 01:23:33 AM |
|
If there were lots of CPPSRB pools then you could engage in hopping to the ones with the smallest backlogs I guess. ![Wink](https://bitcointalk.org/Smileys/default/wink.gif) I'd be interested to see how much you could shorten the expected time to maturity using this approach for an arbitrary number of CPPSRB pools. Let us know how it turns out. ![Wink](https://bitcointalk.org/Smileys/default/wink.gif) Well, that's a much better answer than claiming that the issue doesn't exist! ![Wink](https://bitcointalk.org/Smileys/default/wink.gif)
|
|
|
|
roy7
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 01:28:47 AM Last edit: February 25, 2014, 02:07:30 AM by roy7 |
|
My math-fu isn't at organofcorti's level. I wouldn't even begin to try. ![Wink](https://bitcointalk.org/Smileys/default/wink.gif) Maybe one day Meni will add CPPSRB to his analysis paper.
|
|
|
|
organofcorti
Donator
Legendary
Offline
Activity: 2058
Merit: 1007
Poor impulse control.
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 01:59:08 AM |
|
My math-fu isn't at organofcorti's level. I wouldn't even being to try. ![Wink](https://bitcointalk.org/Smileys/default/wink.gif) Maybe one day Meni will add CPPSRB to his analysis paper. Oh, I figured you'd try simulating something. I did the same for proportional pools a while back, before I knew Meni had already worked on the problem.
|
|
|
|
smooth
Legendary
Offline
Activity: 2968
Merit: 1198
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 02:42:03 AM |
|
If there were lots of CPPSRB pools then you could engage in hopping to the ones with the smallest backlogs I guess. ![Wink](https://bitcointalk.org/Smileys/default/wink.gif) Exactly. The algorithm is not hop proof. Given multiple CPPSRB pools you should hop to the one with the largest buffer, failing that the smallest share log.
|
|
|
|
Biomech
Legendary
Offline
Activity: 1372
Merit: 1022
Anarchy is not chaos.
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 02:52:47 AM |
|
Well, I tried an alt-coin switching pool (tompool) and made less than half. Once again Eligius wins out.
While all the stuff about the payout system is interesting, it has been my real world experience that I make more coin here, even setting a 1 percent donation as automatic. I'll probably still play around some as I go, but at least for bitcoin, Eligius is by far the best pool I've tried. Second was BTCguild, and I make more here.
|
|
|
|
un_ordinateur
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 03:24:14 AM |
|
Man, that was a really interesting read! What about the "lie in wait" attack?
I didn't know that attack until 10 minutes ago. Seriously I've learned things today. Maybe one day Meni will add CPPSRB to his analysis paper.
I believe it is briefly described on page 22; under the name "pay-once-PPLNS". He does not do a thorough analysis of it like the other methods though.
|
|
|
|
organofcorti
Donator
Legendary
Offline
Activity: 2058
Merit: 1007
Poor impulse control.
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 04:45:47 AM |
|
Man, that was a really interesting read! What about the "lie in wait" attack?
I didn't know that attack until 10 minutes ago. Seriously I've learned things today. Maybe one day Meni will add CPPSRB to his analysis paper.
I believe it is briefly described on page 22; under the name "pay-once-PPLNS". He does not do a thorough analysis of it like the other methods though. CPPSRB is closer in nature to SMPPS than PPLNS, and from what I remember he had no plans to add it in.
|
|
|
|
malamashka
Newbie
Offline
Activity: 8
Merit: 0
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 09:10:19 AM |
|
Well, I tried an alt-coin switching pool (tompool) and made less than half. Once again Eligius wins out.
While all the stuff about the payout system is interesting, it has been my real world experience that I make more coin here, even setting a 1 percent donation as automatic. I'll probably still play around some as I go, but at least for bitcoin, Eligius is by far the best pool I've tried. Second was BTCguild, and I make more here.
Yes its great pool, but despite we putting some donations its shame the greatest pool can't keep the stats working a week without a problem ...
|
|
|
|
JLebowskiTheDude
Member
![*](https://bitcointalk.org/Themes/custom1/images/star.gif)
Offline
Activity: 98
Merit: 10
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 10:11:25 AM |
|
I ran simulations of multiple reward systems against real-world data (Eligius shares database/blocks) prior to implementing CPPSRB. (I've rerun this a few times since then, but not recently.) While I admittedly do not know how to boil it down to a single equation, I'm pretty good at simulation ![Smiley](https://bitcointalk.org/Smileys/default/smiley.gif) I took short and long term averages and % deviations from maximum/expected PPS for various time frames accounting for individuals as well as pool-wide earnings with multiple reward systems. While PPLNS (with various different N's) ended up pretty close to CPPSRB's % deviation quickly, CPPSRB always tended to win by at least a few %. Over the longer term and taking into account the real world data (where some miners were not consistent for example) PPLNS left some miners with substantially higher variance than others, while CPPSRB managed to iron this out quite well in most cases. In my simulations CPPSRB kept more miners closer to 100% PPS than PPLNS did over almost all time frames, short and long. -wk This is why many of us are thankful you are involved at the level you are with Eligius. Many many many thanks again for your efforts!
|
|
|
|
Biomech
Legendary
Offline
Activity: 1372
Merit: 1022
Anarchy is not chaos.
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 10:47:05 AM |
|
Well, I tried an alt-coin switching pool (tompool) and made less than half. Once again Eligius wins out.
While all the stuff about the payout system is interesting, it has been my real world experience that I make more coin here, even setting a 1 percent donation as automatic. I'll probably still play around some as I go, but at least for bitcoin, Eligius is by far the best pool I've tried. Second was BTCguild, and I make more here.
Yes its great pool, but despite we putting some donations its shame the greatest pool can't keep the stats working a week without a problem ... Noted, but the coin flows even when the stats page is playing with itself. And it improves a bit every time, it seems. It does get annoying, but I keep coming back. Can't be that bad, apparently.
|
|
|
|
joeventura
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 12:50:31 PM |
|
Website seems to be down this morning with 504s
Shares still being accepted.
|
|
|
|
kodiak1120
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 01:12:43 PM |
|
It's down for me too... hopefully my shares ares still being accepted. I'm not home now so I can't check.
|
|
|
|
Biomech
Legendary
Offline
Activity: 1372
Merit: 1022
Anarchy is not chaos.
|
![](https://bitcointalk.org/Themes/custom1/images/post/xx.gif) |
February 25, 2014, 01:19:33 PM |
|
It's down for me too... hopefully my shares ares still being accepted. I'm not home now so I can't check.
The pool appears to be fine. It's just the stats page again. I'm connected and hashing away.
|
|
|
|
|