Bitcoin Forum

I was trying some different pools this week for analysis on payout totals and came to realize that my PPS pool blew the others away by a significant amount. Proportional and score-based pools whose names I won't mention seemed to max out my reward per round at about 0.3BTC. While I would reach the max relatively quickly, a couple of hours at the most, I found that each round had a tendency to go on for several days at a time. For my hash rate, an average of 0.1BTC per day was not what I was expecting. Then I spent some time in a PPS pool, running the same hardware at the same hash rates and this consistently produced slightly over 1BTC per day. Very impressive, compared to what I was used to getting from other style pools. The PPS pool has a very desirable setup in terms of its fee and the account features it offers.

As far as mining with a prop/score based pool, I can't speak for others with hash rates different from my own, but wouldn't this ring true for everybody? The only way I could think of where these pools would be competitive is if each round finished up within under 8 hours, and, as you all know, the time it takes for a round to finish up is completely unpredictable, even for the bigger pools.

So based on my findings, I'm curious what type of pool works best for others. I'm interested to know if anybody has a similar experience to my own, or if anyone has found their results to be different than the one I wrote about.

Good point about pool hopping. I didn't mention that in my post because I don't understand how to do it and I don't care to know. I assume that pool hopping can be more profitable than pure PPS if you do it right but I'm the type of miner who leaves his miners alone and lets them run 24/7. I just prefer to let them go that way rather than sit and change them around all day. I should also mention that I mine solely for profit, which is the idea that I based my comparison on.

Also, I didn't mean to knock pools with different payouts than pure PPS. Obviously it is important to both the BTC community and the currency itself to have a diverse base of followers and means of generating BTCs.

This is a common misunderstanding. Pool hopping usually is automated. So you just leave the pool hopping software alone and let it run 24/7.

A directory of pools:
 - http://en.bitcoin.it/wiki/Comparison_of_mining_pools

Because you can earn more than that using pool hopping for example.

If you don't hop, I agree that the lowest possible variability is the most desirable option --> PPS is the way to go (by the way: SMPPS etc. are not pure PPS!).

Reasons that might drive one to another pool:
The PPS pool operator is not trustworthy.
The PPS pool is badly connected to the internet and/or the Bitcoin P2P network, resulting in orphan blocks + stale shares.
Lower fees elsewhere.
Some other gimmicks (some people seem to mine at some pools just because the statistics are prettier!)

Pool hopping is direct theft from people who stick with the pool at all times.
I don't know why people actually advocate this as a viable reward method rather than an age-old vulnerability which should be fixed by every pool operator.

Pool hopping is destructive towards the proportional pool & only benefits the person profiting from other people's work.

If everyone practiced hopping the prop. scheme would collapse. Rounds past 43rd percentile probability would never finish & you'd have nowhere to switch to.
Maybe someone ought to release an easy-to-use hopping script or program in public so that the last remaining pools would fix this due to everyone doing it

Yeah, SMPPS etc are better than pure PPS ;)
No, pool hopping is not theft at all, it is simply smart mining. Miners have no obligation to stick to one pool at all times, and have no reason to stick to a pool that is offering to pay them less per-share than its market value. "Pool hopping" is simply working for the highest bidder.

I don't like any PPS derivatives because the EV of mining at one is lower:

if the pool has bad luck you get lower payouts or slower payouts or a combination of both
if the pool has good luck you just get paid normally and the pool operator keeps a bigger buffer - so you don't benefit

Pure PPS is independent of luck. You do not get paid based on blocks. Every single share has a value of 0.00002769 BTc.

According to Eligius 14 day live analysis pure PPS has the highest EV of any payment method.
Every two weeks you are not mining in a pure PPS pool you are losing about 0.5 to 1.5 BTC on average (with 800 mhash of power)

http://eligius.st/~luke-jr/samples/800MH-3/

http://i56.tinypic.com/30mr4vc.jpg

As you can see, 24/7 prop. mining is the worst method in existence.

Since there are 0% pure PPS pools like abcpool.co and btcpool24.com,
I assume miners want to voluntarily lose money to hoppers and inferior payment methods by favouring other types of pools.

Bottom line: Every day you do not mine at pure PPS pools (or exploit prop. pools by hopping) you lose money.

This shit aint logical. Communism is wrong, right?!

I think PPS is better now because of current network conditions. If the network is GROWING in hashrate, you get more rewards as proportional (blocks accelerate in speed, and you acquire more bitcoin). So while you advocate for PPS NOW, it could become a losing choice in the near or far future.

Not only proportional works better when the network hashrate is growing, but also when you join a pool that is experiencing increasing total hashrate power. Part of the problem is that some pools are vulnerable to hopping, making fair proportional miners lose bitcoin, while giving hoppers an advantage. To put it in your words... every day you do not mine at pure pool hopping you lose money.

Someone should track what p2pool is doing in regards to centralized models.

This is purely wrong as it is NOT live data! The example shows an extreme case of bad luck and I can only speculate why Luke forged it that way. Maybe to show the differences in payout systems in a bad luck situation?!

He wouldn't benefit from forging as his own pool uses SMPPS which doesn't do that well in the comparison chart.

It could also be based on what people would earn in *his* pool, were each of those methods were available during that 14 day span.
Of course it isn't live 24/7, in that aspect you are correct. The sample was taken about a month ago. But it's also too precise to be a simulation.

As I said: This is NOT an unmodified sample from any pool.

Why should it be impossible to just simulate a 800 MH/s miner + some randomly distributed blocks by the way? I can do this even in Excel within a few minutes...

http://i52.tinypic.com/116qdf9.jpg
If you hover the mouse over a chart it shows both an estimate, and the actual reward at any time during the 14 day span.
Even if it was simulated it wouldn't make the charts any less valid

Still I would like to hear your explanation why other payment methods are viable when 0% PPS exists.
It's the purest and safest form of mining that exists for the miner.

The Eligius chart shows estimated 24/7 prop. mining earnings nearly 1BTC less per week than PPS [@800mhash] which is significant
(And unlike proportional, hoppers can't benefit from PPS pools, their early shares are treated the same as later shares)

You can simulate not only found blocks but also found shares - then just take different payout models and plot them.

Other payment methods are viable because 0% PPS bears the risk of the pool operator going bankrupt and/or being attacked with a "withholding winning shares" attack, resulting in him loosing money.

Don't you read my posts? The Eligius chart is showing an extreme case of bad luck over a full month! If I presented you the same chart with the opposite amount of luck, all PPS versions/variants would be the bottom line with any other payout system floating on top, as nearly all SMPPS etc. variants are capped at pure PPS level when the pool is lucky.

Why would you mine with anything BUT a PPS pool?

That depends on one's point of view.

A mining analogy. I once lived in a gold mining town.

There were basiaclly 2 types of miners
Type A: worked as an employee of a company, gets paid a wage. -no variance
Type B: prospected on their own or in groups. -high variance

Type:A were divided into full time and casual,
Type:A1 full time worked at the same mine every day for the same pay/hour,
Type:A2 casual moved from mine to mine getting paid a slightly higher rate but didnt get work every day,

Type:B worked in groups or on their own. so
Type:B1 can be solo prospectors
Type:B2 can be group or co-operative prospectors

Type: A1 miners get same $/hour day in day out
Type: A2 miners move from mine to mine at various payrates (usually higher) and get some "days off" with no pay sometimes.

Type: B1 has long periods of no income but when they do find a few nuggets of gold they have a good payday
Type: B2 has shorter periods of no income but as all "finds" are shared on a value_of_find/%work_done they would have smaller more regular payouts than Type: B1

Now each of these miners chose, depending on lots of variables their preffered method of making income - some need stability of income, some could not work in the same hole in the ground on the same machine day in day out, some are looking to "strike it lucky" - who is to say who is right or wrong in thier choice?
I didnt work on the mines, I was a furniture removalist, it suited my lifestyle better, was I wrong?

Are the "casual workers" "evil" for moving between mines and roles? Or are they trying to increase thier income? (not always succeding)

Yes there are some major differences between real gold mining and Bitcoin mining. In Bitcoin mining, income over time can be expected to equalise given 0 fee pools, low stale share rate etc.

In Bitcoin mining
Type:A1 = pool miner on PPS pool
Type:A2 = hopper or B2 on a dry streak
Type:B1 = solo miner
Type: B2 = pool miner on proportional pool
Is how I see it.



Why would you mine on a PPS pool?

When some proportional pools pay 50BTC+TXNs included in block and charge no fee. It is only small now, but already deepbit has had a block with 16BTC txn fees included. On the pool I mine on that would be 66BTC with 0fee shared among miners for that block, how does PPS share value match that?

Actually as TXNs in blocks become large what is the plan for PPS pools to share that? Or is that just a bonus for PPS poolops?

Then there is the PPS pool mining namecoin but paying BTC PPS - in the forum thread the owner has stated he cant afford to pay out a 8,000,000 share round, does this mean he's warning miners to leave at 7,500,000 shares or risk not getting paid for work done?

One pps poolop I know of "hops" his pool to reduce his variance. The pool has only found one block (as of 2 days ago) and that block went invalid. Sustainable?

One PPS variant pool has gone from a positive buffer to a negative buffer in 10 days after 3 larger than 10,000,000 share rounds.
Another pool had 284BTC worth of shares accumulate over a single longblock.

Then there are the PPS variants where poolops will pay some new and some old shares now and some later in more and more complex ways that make it hard for non-mathematician miners know they are getting paid fairly and when.

On proportional at the end of payout there is no "held" funds. No need for a buffer. No risk of not getting paid.No wondering when or if the shars you do today will get paid.
My risk is paying server costs and putting some of my BTC in for rewards.

Some things to consider when choosing a pool:
Have a proven record
Do they communicate important info re sceduled downtime to all members well in advance, or tell them about it after it has happened?
If hardware or other failure is encountered does the pool "dissapear" for a few days or does the poolop tell his IRC chan, post in forums or in other ways inform miners about unexpected downtime as soon as possible?
Honest open operator
Active support by irc, email or forum
Is the pool a registered entity in the country it is trading in? Or the pool operator may be breaking local tax laws.
Is the payout system not only "fair" but transparent and sustainable?
Fee level
Are transaction fees that come with blocks paid to the miners or withheld by the pool?
How are invalid blocks handled by the pool? Wipe submitted shares and start new round, pay for invalids or pay from one valid block to the next valid block thus ignoring invalid blocks and paying for every valid share submitted to the pool?
Does the pool operator do anything to publicise or advance the cause of Bitcoin in the wider community or just run a pool?
Is the poolop planning to change random data in the blockheaders - an if so what to?
Does the poolop actually know how Bitcoin mining works, I saw one poolop claiming to be "building" a block and "it would be finished soon"!
and finally
Is the poolop an anonymous nick?
Does the payout system suit your lifestyle and do you actually understand how it works?


My thoughts

Graeme Tee
Graet
Managing Director
Ozcoin Pooled Mining Pty Ltd
ACN:152 509 272

Member: Organising Committee Bitcoin AU 2011

Actually it would. An example is increasing networks speed and mining in a proportional pool. Since the difficulty remains low until the retarget, you obtain more rounds, thus the PPS reward will always be smaller. This has been observed in a day where the blocks almost doubled in frequency, and a majority of pools got extraordinarily "lucky". Not only do you get bigger rewards if you are in the "lucky" pool with proportional, but the other pools benefited because their rounds were shorter or at least not longer.

Getting paid according to theoretical average PPS, when the network speed is increasing, will result in lower rewards.
Getting paid according to theoretical average PPS, when the network speed is decreasing, will result in better rewards.

I like my earnings to go up when the pool is lucky, and I accept them going down when it's unlucky.

PPS rewards are based on a completely neutral 50% luck, but several other payout schemes also average the same over a long enough period where luck averages out. Prop suffers from unfairness (i.e. not averaging the same as PPS), and I think some *SMPPS variants give miners the shaft when their money buffer has run out and they're getting long rounds.

I do think pool hoppers are a cancer that's leeching off all the non-hoppers. If everyone was a hopper, every prop pool would grind to a halt on the first non-short round. So yeah, I hope prop pools move to other reward schemes.

Since PPS carries a risk for the pool owner, you as a miner risk losing your last payout if the pool goes bankrupt. Granted, that's not a big sum typically - configure your auto-payouts!

PPS is also the most vulnerable to withholding winning shares. If you're a miner, there's really no incentive to submit the winning shares - you keep earning the same no matter what. Sure, a withholder is hurting the pool owner, but he's also helping himself by driving down the network difficulty, however slightly. If that advantage is worth more than the price of 1 share, any sociopath would choose to do it :) But in some other reward schemes, a withholder also hurts himself by preventing his submitted shares from getting paid from the withheld block, so they make withholding unprofitable.

*MPPS pools are not being hopped, your pics are invalid...

I don't advice mining pure PPS pools because sooner or later they will not pay their last round(s) because they were too unlucky...

That's not my pic, that's from Eligius pool owner's chart.
http://eligius.st/~luke-jr/samples/800MH-3/

Also PPS is still not dependent on rounds found.

Go to http://abcpool.co and you will see there is not even a round history in existence.
Miners are getting paid for submitted shares & not found blocks.

He means because bad luck will make the pool owners run out of money and shut down.

I think the reasons why people mine without a PPS pool (especially PROP) are very simple. One part knows what they are doing, another big part probably not, and a small part does not give a shit.

In my opinion it is reasonable to exploit others to get the most benefit for yourself, so I dont really blame anyone who hops pools frequently. If the possibility for exploiting exist you can expect that some people will do it and others will follow.

However, from the view as pool operator it should be primary target to make regular miners happy and give them the best possible value, so the only solution for this is PPS in my opinion (there cant be any better method, and no method is fairer). That's why I have implemented PPS in my pool now too and users can choose if they want to mine with PPS or PROP. Although the payout is not instantly I dont deem the payout scheme to be worse than PROP payout as soon as a block is cleared and credited.

We save all the valid PPS shares (along with a timestamp) in a queue and credit them one after another to the user as soon as the pool balance can allow this. Basically this means every time a block is cleared and credited, the next time when the queue will be checked, it will credit PPS shares to users until there is no more balance left. So it might happen a couple of times, that you cannot cash out your rewards instantly, but on the other hand it can also quite often be that you can cash out your shares before a block has been cleared.

That is not proper PPS though. PPS is supposed to be risk-free for the miner, but from what you describe, the miner is risking not getting paid for ages (or ever) if your pool gets long rounds.

We at rfcpool.com (https://rfcpool.com) used Prop at first and then added PPS too. It started out tied to found blocks, but that was rather unfair so we soon fixed it, so that all PPS shares went instantly to the confirmed balance that could be paid out at any time. But the rounds were long and the pool was incurring losses. The owner paid the miners out of his own wallet, but eventually got tired of the losses and shut it down.

But since we had written the pool website from scratch and it was full of beautiful statistics and stuff, and the servers were robust enough to handle spikes well beyong 200 GH/s (no closing down of registrations for us :P ) - and the users that had to move elsewhere were complaining that the other pools' websites suck and they're unreliable and unsustainable - the owner got strong-armed into reopening :)

So now we're using PPLNS, because it won't bankrupt the pool, it's fair to regular miners (hopper-proof) while being easy enough to understand, and on average miners get the same rewards as on PPS. Everyone is welcome to check us out (https://rfcpool.com) :)

if you are willing to risk one day's worth of payout not a big deal.  Just make sure you get a daily auto payout

That graph is based off of a period of very poor luck.  

Not only that, but there is a difficulty and speed decrease all over the network. Maybe PPS is good when the speed decreases and proportional scores are good when the speed increases.

You don't benefit anything from *other* pools discovering blocks faster & pushing the network 'speed' up

If deepbit gets lucky tomorrow & finds 50% more blocks than normal (hence pushing up the block discovery speed significantly)
your earnings at Slush or any other proportional pool are not going to go up, only deepbit members benefit from it

Pure PPS on the other hand pays always the expected outcome, which is currently 50BTC per 1806098 shares submitted.
There are no variance swings or luck & the size of the pool, network speed etc. is irrelevant

There are many factors to consider when selecting a pool and reward system (in what I consider decreasing order of importance):

• Pool operator trustworthiness (absolutely critical!)
• Pool hopping (applicable to reward systems with a proportional element only)
• Stales and server reliability
• Fees (including withdrawal fees)
• Variance
• Bonuses (find our next block for us and get a bonus 1 BTC!)
• Block reward (50 BTC or 50 BTC + transaction fees - does not apply to PPS)

Pretty much you just have to work out what's best for you.

PPS reduces (but does not eliminate) the need to trust the pool operator and completely kills variance but will usually come with a significant fee.

Solo mining eliminates all problems on this list apart from the rather extreme variance (this is one of my favourite methods).

At this point the best returns would probably come from pool hopping on the largest proportional pool you can (switching to a low fee pps pool).  This will usually give a high expected return and at low variance.  You have to keep your eye on the mining forums though because it's only a matter of time before proportional pools die out.

The worst would probably be mining on a proportional pool without actually hopping but this could be trumped by paying a 10% fee at a PPS pool.

If you have an ethical objection to pool hopping then it's hard to beat solo for expected returns.  Perhaps PPLNS on mineco.in has the highest expected return without the extreme variance of solo mining.

Personally I mine PPLNS at simplecoin.us (I lose block reward transaction fees but gain more in reduced stales).

Oh, why that particular pool? Something the other PPLNS pools don't have?


Their site seems to be down at the moment, so I can't compare stats, but ours has phenominally low stales too, and transaction fees are rewarded to miners  :D

Looks very good.  I only mentioned mineco.in because it was the only pool I knew of (before your post) that did offered PPLNS, 0% fee, and transaction fee rewards (I know they are small right now but this is a big plus in my book).  For me, the grace period for supplying shares for a recently expired round is technically a minus.  Other than that, the list of features is impressive though I would expect to see some mention that there is no fee on BTC withdrawals.  At 25 Gh/s the variance is fairly high but I'm sure it's much better than solo mining for most people.

Given I don't have anything like a complete list of mining pools, I'll rephrase to say that right now the best bet seems to be to find a 0% fee pool using PPLNS and which includes the transaction fee reward.  Avoid PPS unless you've calculated the expected return vs. solo carefully and are happy with the reduction.  Definitely avoid proportional unless you intend to pool hop.

so basically mineco.in

I've started mining at rfcpool which just switched to PPLNS with 0% fee. I've checked out quite a few pools and to be honest, this site looks like it's run by someone that knows what they are doing. The coders hang out in their irc channel, which is a plus.

I personally think that smaller PPS pools are very dangerous to run and mine on, especially if they have a low or zero fee. A malicious individual with access to a competing proportional or the like pool can pull off a bankrupt-to-kill scheme by focusing all shares submitted onto a target PPS pool, but discarding all blocks submitted (this pool would itself by acting as a worker for the PPS pool, but passing the work onto its workers). The losses could be covered up by increasing share stales and marking all blocks as stales, and they would be limited at approximately the PPS pool's fees. The miners would simply feel a long block since the submitted blocks are being marked as stales, but the PPS pool would be paying almost twice as much they earn if the attacker had similar hashrate! Then the attacker withdraws at once all the funds of the PPS pool and resumes normal business, having managed to bankrupt a competing pool.

Well, that's dangerous for the pool - but not a miner.
If you have a high stale rate on a PPS pool (or any pool), just switch to another one - there are enough programs out there to do so unattended even.

Also such attacks can be detected -  and someone directing a significant workload to a small PPS pool will anyways hopefully be tested very fast by the pool owner.
Simply send a share that already solved a block in the past to that miner - if he doesn't submit the valid solution, either ban him right away or retest with another known getwork - then swing the banhammer.

This is a nice idea on first glance but unfortunately totally defeatable.

If the share was solved in the past then it will be in the blockchain. Thus it can be seen that the miner is being tested and he will send it back regardless. A hash table lookup of 140,000 blocks would be pretty trivial.

You would have to find a winning share yourself and NOT make a block out of it, then keep that getwork to one side for future miner tests. And then hope that withholders don't collude and inform each other about your getwork test and add it to their database of ones to always submit back - then you'd need to generate another :)

I'm mining there now as well.  So far I'm getting more BTC/day than I was on BTCguild.  I agree that the site is well done.  Simple yet functional design that supplies me with all of the info I care about.  It's fast & secure and there aren't any ads.  Doesn't seem too popular just yet but I think as more people check it out they'll gain some traction.

It will, as soon as this group of people starts thinking rather than just going with 'whatever is the biggest pool'.


While I do advocate for pure PPS pools (biggest advantage to the miner), I do realize PPLNS is probably 'the' perfect system for new pools.
It needs no startup capital, it's protected against exploiting by hoppers, & it rewards miners fairly.

Give it a year and prop. pools will be history, PPLNS will be the most popular

I'm making more BC mining at Arsbitcoin (PPS Pool) than I was at deepbit. Pretty happy with switching over.

oh, wow, surprise, surprise

Could you explain what you mean by that? I didn't understand it.

@Graet, Very reasonable post about selecting a pool, I second basically everything you said there.

Somehow, it seems that there is a significant number of irrational miners out there and one cannot count on them acting rationally.

Indeed, one can usually count on the majority of miners acting irrationally.

This group of people as mentioned by "wtfman":


Which is probably a majority of miners. When they finally see shrinking earnings & start doing some thinking, PPLNS and PPS pools will explode in popularity.
Or maybe it's began already.

I have observed that to some noticeable degree people with double digit gigahashes being let's say "smart gigahashes" and exhibiting more rational behaviour than, say, majority of "sub gigahash" miners.

Suppose, once one has more or less serious operation going, one has to start calculating yield and fighting for every extra percent of it. As opposed to "lets set it mining while I am not gaming" crowd.

Shrinking margins surely will push evolution forward big time as well.