[0 GH/s 0% fee SMPPS] ArsBitcoin mining pool!

[coblee]

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. 

[RandyFolds]

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.

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?

[John (John K.)]

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 don't see the idea behind your's.

[Meatball]

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.

[being]

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.

[John (John K.)]

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.

[Meatball]

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.

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.

[Big Time Coin]

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.

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."

[Luke-Jr]

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.

[Hotdog453]

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.

[BurningToad]

Well, I think we've created a little too much negative perception ourselves about a negative buffer.  You have to remember that changing to PPLNS won't help us find blocks any faster, it will just distribute the payouts differently when we find them.

A slightly negative buffer has no real ill effect on existing miners.  It is actually mostly just detrimental to new miners looking to join, because they have to help pay miners who already have unpaid credit.

That said, I think I could do a switch to PPLNS sometime in the next 24 hours or so.  Some of the underlying code is already in place.

[Hotdog453]

True, and it's something every pool will go through. This is the first PPS pool I've been in, and the constant stream due to the huge buffer we had in place was pretty sexy.

If I were you, I'd implement the "pseudo" PPLNS, to show what we'd be making if we were on it. It'd give us a nice comparison.

[miscreanity]

Neat! Ars is currently at 2.2% of the entire Bitcoin mining power pool.

http://pident.artefact2.com/

[BurningToad]

Neat! Ars is currently at 2.2% of the entire Bitcoin mining power pool.

http://pident.artefact2.com/

That is because of our shitty luck lately   Based on hashrate, we are a bit higher (3.7%?)

http://bit.ly/hashrate

[miscreanity]

That is because of our shitty luck lately   Based on hashrate, we are a bit higher (3.7%?)

http://bit.ly/hashrate

Even better!

This part at the bottom of the spreadsheet cracked me up:
"Quotes may be delayed up to 20 minutes."

[Big Time Coin]

Well, I think we've created a little too much negative perception ourselves about a negative buffer.  You have to remember that changing to PPLNS won't help us find blocks any faster, it will just distribute the payouts differently when we find them.

A slightly negative buffer has no real ill effect on existing miners.  It is actually mostly just detrimental to new miners looking to join, because they have to help pay miners who already have unpaid credit.

That said, I think I could do a switch to PPLNS sometime in the next 24 hours or so.  Some of the underlying code is already in place.

Just make sure to roll it out at the beginning of a new block, after you have posted a warning on arsbitcoin mainpage that you are about to do it.  A competing pool switched mid-block without sufficient notice recently and it ticked many people off as an unprofessional thing for a pool op to do.

[BadPenny]

Kind of in a pickle here.  Going on 9 hours this block while waiting on a ~10,000,000-share block to confirm to replenish the depleted buffer.

[RandyFolds]

Kind of in a pickle here.  Going on 9 hours this block while waiting on a ~10,000,000-share block to confirm to replenish the depleted buffer.

Probability has been slapping us around a bit. It's no fun...

[BurningToad]

So, I said that I could try and do PPLNS within 24 hours, but that isn't going to happen.  It turns out there are too many code changes that need to happen and that need to be tested.  The actual PPLNS calculation code is the easy part.  I can easily perform PPLNS payment calculations based on when we find a block now on ArsBitcoin.  However, making sure that everything is 100% correct, making sure we handle blocks that go invalid, showing unconfirmed rewards, confirmed rewards, PPLNS round history, personal user stats that work with our existing stats, etc, is all proving to take longer than anticipated to implement.

On top of that, I have some important deadlines at work tomorrow, and I just can't spend any more time on this right now.

So, now Ars becomes a grand SMPPS experiment as we venture back into unlucky negative buffer territory (which is what we started with, btw, as we went negative the first time a share was submitted and it didn't find a block.)

Hopefully our luck turns soon, but if not, I thought I would again explain how payouts work when we can not pay all of the Total PPS work.

So, various miners have a Total PPS work > Paid PPS Work.  This is because the pool doesn't have BTC to pay anyone.  Once we get more BTC, the following occurs:

Each miner has an 'ideal payment' of Total PPS Work - Paid PPS work.  For example, Miner 1 could have 100 Total PPS Work, 99 Paid.  His ideal payment is 1 BTC.  Miner 2 has 50 Total PPS work, and 48 Paid PPS work, or an ideal payment of 2.

Lets say the pool just confirmed a new block that is worth 1 BTC.  First, we get the ratio of what we can pay everyone.  Total outstanding ideal payments in this 2 miner pool is 3 (2 + 1.)  So, our total ratio that we can pay people is (1 / 3) or 0.333333etc.  

This means that Miner 1 gets (1 * 0.333333etc) BTC and Miner 2 gets (2 * 0.3333333etc) BTC, for a total of 1 paid out (all that the pool has available.)

Note that this isn't exactly like Eligius' SMPPS, where older shares get paid first I believe.  It also isn't exactly like ESMPPS or RSMPPS or whatever.  It pays all miners equally based on their 'ideal pay' when the pool has BTC available to pay.  If someone stops mining and still has some 'ideal pay' amount, it will slowly get paid as the pool gets BTC.  However, people who DO stay will obviously be building MORE 'ideal pay', so they can be sure that they are still getting paid a much higher share of the pools BTC than anyone who stopped mining.

So, in summary.  I still plan to work on PPLNS in the near future, but it will not be done in the next 24 hours.  Our bad luck seems to be continuing, so we will most likely not be able to pay 100% PPS value instantly like we have been able to for the past month+.  However, the pool will obviously still be solving blocks and still pay out as much as it can, just like any other pool out there.  

If time goes on and we still have a negative buffer in the future, I still think we should switch to PPLNS.  What this means is that, at that time, it would no longer be possible to make Total PPS work = Paid PPS work.  Just like if you mine at any (non straight PPS) pool where the pool luck is < 100% over the time you mine there, you will not be able to get paid 100%.  

Sorry that the luck party has decided to visit others for a while (damn Eligius and their 900 BTC+ buffer like we used to have  ).

[Luke-Jr]

Hopefully our luck turns soon, but if not, I thought I would again explain how payouts work when we can not pay all of the Total PPS work.

So, various miners have a Total PPS work > Paid PPS Work.  This is because the pool doesn't have BTC to pay anyone.  Once we get more BTC, the following occurs:

Each miner has an 'ideal payment' of Total PPS Work - Paid PPS work.  For example, Miner 1 could have 100 Total PPS Work, 99 Paid.  His ideal payment is 1 BTC.  Miner 2 has 50 Total PPS work, and 48 Paid PPS work, or an ideal payment of 2.

Lets say the pool just confirmed a new block that is worth 1 BTC.  First, we get the ratio of what we can pay everyone.  Total outstanding ideal payments in this 2 miner pool is 3 (2 + 1.)  So, our total ratio that we can pay people is (1 / 3) or 0.333333etc. 

This means that Miner 1 gets (1 * 0.333333etc) BTC and Miner 2 gets (2 * 0.3333333etc) BTC, for a total of 1 paid out (all that the pool has available.)

Note that this isn't exactly like Eligius' SMPPS, where older shares get paid first I believe.

No, this is exactly the same as Eligius' SMPPS-- we do not prioritize (nor track) any shares.