Yee haw, guessing I can finally post here now.
From my point of view: Once we dip into the negative balance, people will drop like flies. It's just normal. I moved to ARS because, at the time, deepbit was getting DDOSed insanely hard, and seeing my boxes just sit there doing nothing is a terrible, terrible thing.
Looks like we will probably hit negative sometime today, eh? Looking at ~60 or so BTC balance right now.
Do you have any time frame for the change over to PPLNS? Or at least a timeframe for getting the pseudo-code written, so we can see what we'd be getting if we were on it?
I like the ARS pool since the owner is obviously intelligent and responsive, but it's a house of cards; once the payouts stop or slow, people running larger farms are going to bail.
Topic: [0 GH/s 0% fee SMPPS] ArsBitcoin mining pool! - page 31. (Read 123750 times)
Actually, if everyone stopped mining when the buffer was negative, there would be btc earned that would simply not be paid at all under the current PPS model, even after all mined blocks had been confirmed. 
Not correct. SMPPS never rewards more than it has. You might have people left with "extra credit", but that is not ever guaranteed to be worth anything. 
Has anyone actually done any maths about the negative buffer? Like how long in reality would the delay be?
I see a lot of people crying about it, but it seems as if many of them don't really understand what would the negative buffer really mean. It's not like you would get less btc for the work done. It would only mean, that the pay would be delayed a bit. Even if the buffer stays constantly negative, you would still get paid, when new blocks are found. But with a tiny delay.
I see a lot of people crying about it, but it seems as if many of them don't really understand what would the negative buffer really mean. It's not like you would get less btc for the work done. It would only mean, that the pay would be delayed a bit. Even if the buffer stays constantly negative, you would still get paid, when new blocks are found. But with a tiny delay.
The delay would vary depending on when the pool found it's next block and how quickly it was confirmed. So, for sake of argument let's say we hit 0 BTC buffer and everyone is paid. We then start working on the next block, building up shares we're waiting to be paid for. Let's say it takes 12 hours to find a block. We then have to wait for that block to get 120 confirmations before that 50 BTC can be released for payouts. Of course, during the time we're waiting for approval, we all keep building up more shares that are waiting to get paid. Even if you stop, you will eventually get paid for all those shares, assuming the pool doesn't go under.
Hopefully we'd find more blocks and work our way out of this, but it's not really a simple calculation to determine delay in payment.
Actually, if everyone stopped mining when the buffer was negative, there would be btc earned that would simply not be paid at all under the current PPS model, even after all mined blocks had been confirmed.
But if everyone continues mining, the pay could be delayed indefinitely. If you wanted to know a certain timeframe, the best you could do is a probability simulation that the buffer would return to non-negative value within that timeframe. You would have a % chance of return to positive, which would mean the pool got caught up on payments. The math is not easy.
from a blackjack website: "Risk of ruin problems are mathematically usually very complicated. It is easier and more convincing to run a random simulation instead."
Has anyone actually done any maths about the negative buffer? Like how long in reality would the delay be?
I see a lot of people crying about it, but it seems as if many of them don't really understand what would the negative buffer really mean. It's not like you would get less btc for the work done. It would only mean, that the pay would be delayed a bit. Even if the buffer stays constantly negative, you would still get paid, when new blocks are found. But with a tiny delay.
I see a lot of people crying about it, but it seems as if many of them don't really understand what would the negative buffer really mean. It's not like you would get less btc for the work done. It would only mean, that the pay would be delayed a bit. Even if the buffer stays constantly negative, you would still get paid, when new blocks are found. But with a tiny delay.
The delay would vary depending on when the pool found it's next block and how quickly it was confirmed. So, for sake of argument let's say we hit 0 BTC buffer and everyone is paid. We then start working on the next block, building up shares we're waiting to be paid for. Let's say it takes 12 hours to find a block. We then have to wait for that block to get 120 confirmations before that 50 BTC can be released for payouts. Of course, during the time we're waiting for approval, we all keep building up more shares that are waiting to get paid. Even if you stop, you will eventually get paid for all those shares, assuming the pool doesn't go under.
Hopefully we'd find more blocks and work our way out of this, but it's not really a simple calculation to determine delay in payment.
Unless the negative buffer is so huge that people gets paid out really slow, I doubt that would be a problem. Worse case is ending up like Eligius now; payments more then 0.33 gets put on a not so long waiting list.
Has anyone actually done any maths about the negative buffer? Like how long in reality would the delay be?
I see a lot of people crying about it, but it seems as if many of them don't really understand what would the negative buffer really mean. It's not like you would get less btc for the work done. It would only mean, that the pay would be delayed a bit. Even if the buffer stays constantly negative, you would still get paid, when new blocks are found. But with a tiny delay.
I see a lot of people crying about it, but it seems as if many of them don't really understand what would the negative buffer really mean. It's not like you would get less btc for the work done. It would only mean, that the pay would be delayed a bit. Even if the buffer stays constantly negative, you would still get paid, when new blocks are found. But with a tiny delay.
Everyone here is entitled to their own opinions, and it's great to see dialog on the topic for both sides. What it comes down to though is BT and what he wants to do. It's his pool and if he wants to make the switch, so be it. If people leave or people join because of it, so be it. There's tons of options out there and people should use the pool that works the best for them. That's what's awesome about all these pools, there's so many options and if there's not one you like, you can easily start up your own.
BT's pool has been the best value because of the 0% fee and the immediate payout (due to the positive buffer), but we're all kidding ourselves if we think that's realistic for the long term. First, if everyone set their donations to 0%, BT would shut down in a month or two just due to the expense of running a server out of pocket (which I believe BT said at one point was like $7-$9/day). I know I couldn't afford to pay a few hundred dollars a month for the privilege of letting people mine in my own pool. Second, we're about to see the effects of 'negative' buffer. We've already started to hit the point where payments are getting delayed because we don't have the 'confirmed' buffer available. At the point we're truly negative I think you'll see a lot of folks jump ship. We had a 500-800 BTC positive buffer for a month or two. How many people would stick around with a 500-800 negative buffer for a month or two?
I think PPS has been a great experiment and it's been nice to have that rock steady payout, day after day, but it's not sustainable and at some point BT will need to close up shop or switch to a new system. It's not fair for us to expect him to keep hundreds of BTC secure when we're positive and be on the hook for hundreds of BTC when we flip negative.
BT's pool has been the best value because of the 0% fee and the immediate payout (due to the positive buffer), but we're all kidding ourselves if we think that's realistic for the long term. First, if everyone set their donations to 0%, BT would shut down in a month or two just due to the expense of running a server out of pocket (which I believe BT said at one point was like $7-$9/day). I know I couldn't afford to pay a few hundred dollars a month for the privilege of letting people mine in my own pool. Second, we're about to see the effects of 'negative' buffer. We've already started to hit the point where payments are getting delayed because we don't have the 'confirmed' buffer available. At the point we're truly negative I think you'll see a lot of folks jump ship. We had a 500-800 BTC positive buffer for a month or two. How many people would stick around with a 500-800 negative buffer for a month or two?
I think PPS has been a great experiment and it's been nice to have that rock steady payout, day after day, but it's not sustainable and at some point BT will need to close up shop or switch to a new system. It's not fair for us to expect him to keep hundreds of BTC secure when we're positive and be on the hook for hundreds of BTC when we flip negative.
Thanks for the links Luke, much needed for the discussion.
I wrote a whole spiel, then realized I was volunteering sage business advice to strangers, which is stupid, so I e-mailed it to myself instead to put in the idea locker.
Good luck with your pool's problem of when the buffer runs out.
I wrote a whole spiel, then realized I was volunteering sage business advice to strangers, which is stupid, so I e-mailed it to myself instead to put in the idea locker.
Good luck with your pool's problem of when the buffer runs out.
I don't see the idea behind your's.
Thanks for the links Luke, much needed for the discussion.
I wrote a whole spiel, then realized I was volunteering sage business advice to strangers, which is stupid, so I e-mailed it to myself instead to put in the idea locker.
Good luck with your pool's problem of when the buffer runs out.
I wrote a whole spiel, then realized I was volunteering sage business advice to strangers, which is stupid, so I e-mailed it to myself instead to put in the idea locker.
Good luck with your pool's problem of when the buffer runs out.
You don't know how to google simple phrases that you come across, but don't understand and yet you are a sage...sure...cough-17-year-old-in-a-basement-cough.
Can I pay you some BTC to post your sage advice publicly? I am completely serious. What's the price to show us your brilliant expose on payout schemes?
I wrote a whole spiel, then realized I was volunteering sage business advice to strangers, which is stupid, so I e-mailed it to myself instead to put in the idea locker.
Gee, thanks.
Thanks for the links Luke, much needed for the discussion.
I wrote a whole spiel, then realized I was volunteering sage business advice to strangers, which is stupid, so I e-mailed it to myself instead to put in the idea locker.
Good luck with your pool's problem of when the buffer runs out.
I wrote a whole spiel, then realized I was volunteering sage business advice to strangers, which is stupid, so I e-mailed it to myself instead to put in the idea locker.
Good luck with your pool's problem of when the buffer runs out.
Toad -
The current PPS method is the reason we mine at your pool, and I know that many or most who mine here are only here because of the perfect 1 to 1 expected value, with 0% fee. In other words, the way you are doing it right now, with the option for PPS, makes you the highest EV pool available 24/7, which is why you have seen the recent growth. Changing the payout system, with no option for people to keep the current payout system, will force me out of the pool, and I predict 50%+ of your hashrate will go with me.
To reiterate: your PPS payout system is perfect and cannot be improved in any way. It ain't broke, so don't waste time trying to fix it.
You are scaring us with this other payout system talk, it's either Score, PPS, or Prop. Once you start adding letters in there you are just going to confuse everyone. At least tell us what the system is called and briefly explain it or give a link in the post where you announce you are strongly considering changing to the system.
- The Dirty Dozen
The current PPS method is the reason we mine at your pool, and I know that many or most who mine here are only here because of the perfect 1 to 1 expected value, with 0% fee. In other words, the way you are doing it right now, with the option for PPS, makes you the highest EV pool available 24/7, which is why you have seen the recent growth. Changing the payout system, with no option for people to keep the current payout system, will force me out of the pool, and I predict 50%+ of your hashrate will go with me.
To reiterate: your PPS payout system is perfect and cannot be improved in any way. It ain't broke, so don't waste time trying to fix it.
You are scaring us with this other payout system talk, it's either Score, PPS, or Prop. Once you start adding letters in there you are just going to confuse everyone. At least tell us what the system is called and briefly explain it or give a link in the post where you announce you are strongly considering changing to the system.
- The Dirty Dozen
I think that the switch needs to be made.
SMPPS is broken, no doubt about it. At first, I was intimidated by Aexoden's constant talk of PPLNS, but I have come to see the light after some thorough explanations from my own research and from the guys on the IRC. We have spend the last two months with a massive positive buffer. No one would be here if the buffer was negative 1000btc, hell, I am not sure I would have stuck around and I've been here a while.
Everyone joined in a heartbeat when it was positive, since there is no variance for the 24/7 miners and it was the most profitable place to be if your hopper didn't like the odds at the Prop pools. Why serve as the backup plan for the hoppers and deplete the buffer for the 24/7 miners? Going seriously negative would literally be the death of the pool. Also, as someone mentioned earlier, the liability of having to hold and secure $10,000+ worth of other people's money is a big one...
50% of the hashrate will bail because a more effective and fair system is instituted? There are a line of people waiting for a big PPLNS pool to join, and the biggest one in existence is undergoing explosive growth.
To reiterate: your PPS payout system is perfect and cannot be improved in any way. It ain't broke, so don't waste time trying to fix it.
FWIW, Eligius (the original SMPPS pool) also agrees with BurningToad's conclusion that SMPPS is inherently broken (though not necessarily the reasons), and is leaning toward a PPLNS-like system as well. Bitp.it also considered SMPPS and decided against it (though they did end up using the ESMPPS variant). Basically, every pool operator who has considered SMPPS is in agreement that it is "broke" in the long run, and needs to be fixed in some way or another. If you run a lot of mining rigs, I would encourage you to become familiar with the various reward systems in more detail.Here are some helpful links:
http://eligius.st/wiki/index.php/Maximum_PPS
http://eligius.st/wiki/index.php/Shared_Maximum_PPS
http://eligius.st/wiki/index.php/Capped_PPS
http://eligius.st/wiki/index.php/Pay_Per_Last_N_Shares
http://eligius.st/~luke-jr/samples/800MH-3/
Toad -
The current PPS method is the reason we mine at your pool, and I know that many or most who mine here are only here because of the perfect 1 to 1 expected value, with 0% fee. In other words, the way you are doing it right now, with the option for PPS, makes you the highest EV pool available 24/7, which is why you have seen the recent growth. Changing the payout system, with no option for people to keep the current payout system, will force me out of the pool, and I predict 50%+ of your hashrate will go with me.
To reiterate: your PPS payout system is perfect and cannot be improved in any way. It ain't broke, so don't waste time trying to fix it.
You are scaring us with this other payout system talk, it's either Score, PPS, or Prop. Once you start adding letters in there you are just going to confuse everyone. At least tell us what the system is called and briefly explain it or give a link in the post where you announce you are strongly considering changing to the system.
- The Dirty Dozen
The current PPS method is the reason we mine at your pool, and I know that many or most who mine here are only here because of the perfect 1 to 1 expected value, with 0% fee. In other words, the way you are doing it right now, with the option for PPS, makes you the highest EV pool available 24/7, which is why you have seen the recent growth. Changing the payout system, with no option for people to keep the current payout system, will force me out of the pool, and I predict 50%+ of your hashrate will go with me.
To reiterate: your PPS payout system is perfect and cannot be improved in any way. It ain't broke, so don't waste time trying to fix it.
You are scaring us with this other payout system talk, it's either Score, PPS, or Prop. Once you start adding letters in there you are just going to confuse everyone. At least tell us what the system is called and briefly explain it or give a link in the post where you announce you are strongly considering changing to the system.
- The Dirty Dozen
I'm all for switching to PPLNS as well. I know I wouldn't want to be responsible for a huge wallet full of buffer BTC. As of the switching while negative, there's a simple solution to that, don't switch while negative.
If/when the decision is made to switch over, if we're negative, just hold out until we're positive again. Of course, that assumes we're not like 1000 BTC negative, but if it's not much, I'm sure waiting would be fine.
The other option is to switch asap while we still are positive.
If/when the decision is made to switch over, if we're negative, just hold out until we're positive again. Of course, that assumes we're not like 1000 BTC negative, but if it's not much, I'm sure waiting would be fine.
The other option is to switch asap while we still are positive.
I agree with Meatball. Additionally, does anyone else picture his avatar speaking when they read his posts?
I'm all for switching to PPLNS as well. I know I wouldn't want to be responsible for a huge wallet full of buffer BTC. As of the switching while negative, there's a simple solution to that, don't switch while negative.
If/when the decision is made to switch over, if we're negative, just hold out until we're positive again. Of course, that assumes we're not like 1000 BTC negative, but if it's not much, I'm sure waiting would be fine.
The other option is to switch asap while we still are positive.
If/when the decision is made to switch over, if we're negative, just hold out until we're positive again. Of course, that assumes we're not like 1000 BTC negative, but if it's not much, I'm sure waiting would be fine.
The other option is to switch asap while we still are positive.
I've mentioned in IRC that I am considering a switch to a PPLNS payment system. This has a few advantages over our current SMPPS system.
I'm personally in full support of ultimately making a change to PPLNS. SMPPS has its advantages, but I believe that PPLNS is better in the long run.
However, when you implement it, do be careful about handling difficulty changes correctly. A naive solution can cause various shares around the difficulty change to have different expected values, as you can see in the following graph, which shows the results of a couple of solutions around a hypothetical difficulty increase from 100 to 200, where N is the difficulty:
http://www.calindora.com/tmp/pplns.png
(PPLNS refers to always taking the last N shares on payout. This means you have to always be keeping extra so you have enough in case a block is found shortly after a difficulty change. PPLLNS is a variation that expands N from old_difficulty to new_difficulty as shares come in. Neither has the correct expected value per share around the difficulty change. The best solution (as far as I know) is Meni's variation, but I apparently don't know it well enough in my head to spit out here. I'd have to dig it up from wherever I read about it.)
Another thing to consider is that if and when transaction fees become a significant part of the block reward, pools are going to be hard-pressed to continue to keep them entirely as a fee. Adding those seems comparatively easy with PPLNS, while SMPPS poses more of a challenge. (I don't think variable block rewards do anything weird to the expected value per share, but I could be wrong.)
If we get in a negative buffer, i'm not sure how to handle it yet. It we switched to pure PPLNS, the only way to really handle it is to basically say that we aren't going to pay out the rest of the unpaid work in SMPPS.
Keep in mind, I'm not too familiar with PPLNS or the discussions surrounding the differing methods in general.
What about holding a few percent as a constant buffer? Is that like CPPSRB? Keeping somewhere between 1-5% on reserve should cover decently. The specific rate should be able to cover a reasonable payout, ideally for the maximum historical duration without finding a block.
Hello,
I've mentioned in IRC that I am considering a switch to a PPLNS payment system. This has a few advantages over our current SMPPS system.
1) The pool can't go deeply negative which may cause some people to leave for greener pastures.
2) It doesn't require the pool to hold on to a large SMPPS buffer of BTC, putting more of the responsibility of holding BTC back in the miner's hands.
3) Like #2, it means when the pool is lucky, you get paid for that luck immediately instead of the pool holding on to it to pay you later.
4) Obviously this means payout variance will increase a lot (lately the variance has been about 0 with a positive buffer.) However, it is still a fair payment system and because of the reasons mentioned here, I think it is a better system in the long run.
My plan is to try and code it as soon as I can, and then basically enable it side-by-side with the current SMPPS system. This means that I will show you "simulated" payout stats on the site for PPLNS so that we can test it and see how it works before actually using it to pay anyone. If we see some problem or decide not to use it, we don't have to do so.
If we switch systems with a positive SMPPS buffer, that buffer will be distributed to miners based on all submitted SMPPS shares.
If we get in a negative buffer, i'm not sure how to handle it yet. It we switched to pure PPLNS, the only way to really handle it is to basically say that we aren't going to pay out the rest of the unpaid work in SMPPS. However, if you keep mining with Ars under PPLNS, then you would basically make that buffer back the next time we are lucky, just like the SMPPS system would have. However, this means that those who left the pool for greener pastures would not get their full PPS work paid. There is no perfect solution to this problem, which is a big part of the reason that we should consider switching in the first place.
I've mentioned in IRC that I am considering a switch to a PPLNS payment system. This has a few advantages over our current SMPPS system.
1) The pool can't go deeply negative which may cause some people to leave for greener pastures.
2) It doesn't require the pool to hold on to a large SMPPS buffer of BTC, putting more of the responsibility of holding BTC back in the miner's hands.
3) Like #2, it means when the pool is lucky, you get paid for that luck immediately instead of the pool holding on to it to pay you later.
4) Obviously this means payout variance will increase a lot (lately the variance has been about 0 with a positive buffer.) However, it is still a fair payment system and because of the reasons mentioned here, I think it is a better system in the long run.
My plan is to try and code it as soon as I can, and then basically enable it side-by-side with the current SMPPS system. This means that I will show you "simulated" payout stats on the site for PPLNS so that we can test it and see how it works before actually using it to pay anyone. If we see some problem or decide not to use it, we don't have to do so.
If we switch systems with a positive SMPPS buffer, that buffer will be distributed to miners based on all submitted SMPPS shares.
If we get in a negative buffer, i'm not sure how to handle it yet. It we switched to pure PPLNS, the only way to really handle it is to basically say that we aren't going to pay out the rest of the unpaid work in SMPPS. However, if you keep mining with Ars under PPLNS, then you would basically make that buffer back the next time we are lucky, just like the SMPPS system would have. However, this means that those who left the pool for greener pastures would not get their full PPS work paid. There is no perfect solution to this problem, which is a big part of the reason that we should consider switching in the first place.
I think those numbers are overall BTC amounts and not percentages. Please correct me if I'm wrong.
Pretty sure that BT has said that people that donate more than the current Weighted Donation % will have their names bolded. Right now the weighted percentage is .487% and 21 people of the 40 are bolded.
Jump to: