Topic: Deepbit PPS vs. smaller Pools (Read 4824 times)

Ninjacoin.com is offering PPS at a 3% fee, and btcpool24.com offering PPS at 0% fee.

0% PPS does seem unsustainable, but there is little risk to the miner because there is no minimum payout with btcpool24.  Not sure about ninjacoin.

ArsBitcoin uses SMPPS, not PPS. SMPPS is a bad method.

Check out ARSbitcoin 0% fees (I expect this will be set to 1% soon to cover cost of scaling though)

They - or we  - have been very lucky and currently are floating a 800+ BTC buffer to keep those bitcoins pouring in even in the event of a "block from hell"

currently getting 0.000029570125 per share as opposed to 0.00002661294865101 at deepbit.

No payout algorithm can see your hash rate, you will always be evaluated at your sent shares rate.

It is true however that in PPLNS if you send 1000 Shares in 1 minute and then 1 share 1000 minutes later it might be the case that you only get paid for that single share.

I don't know of mining programs that display your current luck, but it should be within limits and even out anyways, so mining luck is not so much of a concern to me with PPLNS.

Also if N is reasonably big (the biggest useful "N" is btw. 5 000 000 000, meaning each share will be worth one Satoshi at each payout, if you don't pay out transaction fees) this should be less of an issue.

PPLNS punishes your luck, there is a luck involved when sending shares, you don't get paid based on hashrate your miner shows, but based on sent shares - you can see with latest poclbm what your lucky hashrate looks server-side, if you have guiminer it also can show server-side hashrate

before going offline x8s.de showed estimated reward on Prop.P. and PPLNS, as I had 3 accounts, 2 of those mined 24/7, I saw one was +2.5% over PP the other was -3% under PP, and there goes "fairness" of PPLNS - if your miner is unlucky @ end of the round you get underpaid

No. With hopping-proof methods (geometric, PPLNS) the expected payout per share is the same no matter when it was submitted.

How does having a CPU or other weak miner always mining for you under your account while pool hopping with your primary miner get handled by the various schemes? It would seem PPS wouldn't be affected at all but these others might be skewed a bit if you dumped a lot of hashing power early on then moved on to another pool while leaving say 20MH at the original pool. Yes/No?

Of course it's not the only risk. But the risk due to to variance can be calculated and planned for.

Could work, but it will probably have a lot of false positives, and what stops the attacker from creating a new account once in a while?

you would end up banning someone running a back-up pool where 300Mh/s are going towards the main pool and 3Mh/s are going to the backup
or some poor CPU miner

rfcpool.com does 7% fee PPS

The only remaining question for me is then:

Is block withholding (which essentially removes the hash rate of the withholder) the only risk left in a "pure" PPS pool?

It should be detectable by the way, if you are willing to invest a few coins for that (or cheat a tiny bit on your miners) - as soon as someone has solved a valid block you send out that getwork manually to all of your miners connected at that time and kickban everyone that doesn't send a reply. If they send a reply back, you can either choose to pay them for their honesty or just treat this share as stale and not pay it.

All in all miners loose ~3000 Satoshis on that share but you gain a lot more security without having to beg all mining client developers to implement oblivious shares (as far as I understood it, it also requires changes on the miner's side).

On the other hand, with the current pool model you won't be competitive anyways, there are already enough other pools out there - so you anyways need to be more innovative.
As an example:
A miner can also deployed via BOINC (which is also able to handle requesting work and managing credits etc.) and some exchanges can be scripted to automatically change money to USD/EUR... combine this, put in a reasonable fee and get some profit going on. You wouldn't even need to handle payouts, as both MtGox allow you to move BTC (TH) or BTC/USD (MTG) between accounts, you just move the mined money to their account (and if they hand you an API key, automatically bot-trade that money for fiat, if desired) and let them handle it from there. You might even be able to get lower fees + referral income (at least on TH) and all of that can be done _right now_ already if you combine the right pieces.

I stand corrected, didn't know you had the method live already. Only read a while back that you would be adding it 'soon'.

Of course, any rational person (with enough trust in your BTC reserves) will be mining PPS at your pool instead of deepbit

You did, but you also said "there is no need to pay any fees (10% or 7%) for PPS". No point arguing this though.

You stop mining for the PPS pool when it goes bankrupt, which is exactly what you should do when an *MPPS pool goes red. The difference is that a PPS operator takes the fee required to prevent this from happening. In case of bankruptcy, if the operator is honest and takes the necessary planning, you will not lose payment for work already submitted.

If a PPS operator wants to eliminate block withholding risk, he can lobby for my oblivious shares proposal.

Actually I believe I said "PPS variant".


You are correct, there are sometimes delayed payments, but not to the extent that SMPPS or other *PPS variants offer. For that matter, even true PPS could pay less than 100%, notably if the pool operator goes bankrupt due to block withholdings. The point of ESMPPS is to offer payment to those who deserve it the most, that is the way the reward scheduling algorithm works.

?MPPS is not PPS so don't mislead people by presenting it as such.

ESMPPS has the same fundamental problems as SMPPS. If the balance is negative you will receive a low reward short-term. And since everyone will leave there's not going to be a long-term.

Just so everyone knows, there is no need to pay any fees (10% or 7%) for PPS. We just started offering a 0% PPS variant that rewards everyone fairly for their time.


That is exactly why we went with ESMPPS instead of SMPPS. Unlike SMPPS, ESMPPS does not stop paying miners when the reserves get low. In fact, to a large extent, ESMPPS was designed with the idea that negative reserves are an inevitable.

Please stop spreading this mis-information. You've even posted in our thread so you're aware we exist and offer PPS, however new we are.

To clarify - rfcpool offer PPS at 7% fee. Sure we're new, and small, and probably untrusted by most, but that doesn't make your statement any more true.

'Disarm' is an even better word. Hoppers are welcome to hop, but they won't receive more than normal (unlike proportional) but also not less.

As an added bonus, if they don't take a fee to boost the reserve, with probability 1 it will eventually go too much in the red. Whether people will be smart enough to leave then is a different question.

It does, pool hoppers. Or maybe 'deter' is a better word. Nobody has an incentive to jump onto pools like mineco.in at any 'certain' time to gain a statistical advantage.


Fair enough, I'll be first to admit I would run if the balance went too much in the red. If enough people did this the SMPPS pool would die from a lack of miners and thus blocks towards paying off the 'debt'.

Good for you, just make sure to jump ship when their balance becomes negative and you won't get anywhere near that.

That's a myth, PPLNS doesn't punish anyone, it just gives everyone a fair payout in exact proportion to their work (on average).

Did they decide what to write on the tombstone yet? You've got to plan ahead when you're committing suicide.

Unless it's not really 0% fee or not really PPS.

Do you have a link? MtRed would become one of the most popular pools if that were true.
Nobody can compete with 0% PPS.


I've mined at BTCguild for ages, it's just another proportional pool which has all the problems of pool hopping. The 0% fee is irrelevant because of the flawed distribution model of rewards.

PPS means you get paid for every single share you submit (fixed amount). Other pools have modified PPS schemes that are not the same thing (SMPPS, PPLNS)
No pool hopping penalties, 24/7 miner rewards, or incentive to jump in during the beginning of rounds, just getting paid for what you contribute. Pure PPS is effectively 0 variance over any time period.

The problem being that you need quite a lot of BTC in buffer to pull it off, in case of bad luck streaks and very long rounds.
Deepbit is the only pool which offers pure PPS rewards, at a hefty fee (10%)

BTC Guild uses a Proportional Payment Method, not PPS (Pay per share)

JackOfDiamonds,

btcguild is PPS and has not a fixed fee, so it should be the best one for casual and not so casual miners, did you look into it?

spiccioli.

Yeah I know, the 10% fee is a lot but MtRed is coming out with 0% PPS in a few days because of all the pool hopping problems they've been having lately.

Yes, the PPLNS payment method rewards 24/7 miners and punishes people who disconnect during a round (accidental, or pool hopping, it doesn't know or care)

Slush has something similar, a proprietary scoring system.
The longer a round gets the more valuable new shares become, which solves the pool hopping problem & gives nobody an incentive to hop on slush in hopes of exploiting early blocks.

PPLNS is also foolproof and can be adapted by anyone because you don't need a massive balance of bitcoins for a 'rainy day' (pure PPS) or even a medium sized buffer (shared maximum PPS).
I think it will become a success among all pools in the future, or people will simply join pools with PPLNS after seeing decreased avg. earnings in the long run mining in proportional pools infested by hoppers

Of course pure PPS is 'ideal' for both the 2-hour casual miner and the 24/7 'pro' but since only the biggest pools with huge earnings like deepbit can offer it, it's not a realistic option.
Even they take a 10% fee because you gain zero variance in return. That's steep for big miners, and a lot of lost cash for 'certainty'.

I've been using Eligius as well for some time now. Is mineco.in better if you mine 24/7?

This

"The long run" meaning months and years, not a few days or weeks..

You can't get 'consistently' more than 100% of your expected earnings over the long run from prop. pools unless you cheat by exploiting defective pools with no scoring methods (pool hopping)

Deepbit is large enough for their sample to be taken seriously, after ½ a year at multi ghash speeds, their luck pretty much averages to +-0% if averaging all past difficulties
https://www.deepbit.net/stats

The only pools I get consistently 140% to 200% of my expected earnings in the long run are proportional pools. Makes you think, eh?

The only pool where I consistently get 99-100% of my expected earnings in the long run is Eligius with their SMPPS system. I'm constantly leaving prop. pools for other types due to the massive variance that occurs at current difficulties + losses from pool hoppers

I'm also looking into Mineco.in right now which has a PPLNS payment scheme,
far as I can tell from reviewing the idea for a few days, it's better than any other system in existence right now

You can fetch getworks from any pool in advance and just present the one you want to have solved to the miner locally. This means you'll have even less lag than a miner who fetches getworks after submission of results!

All in all, you don't loose any time by switching pools.

You might lose some shares from reconnecting but the difference due to the proportionate weight of shares in a short run more than makes up for it if the pool hopper gets in at the right time.

After all, the time needed for 1 reconnect probably won't lose you more than 1 share per GHash.

This isn't you saying that pool hopping doesn't work, is it? There's so much data out there proving it does - both empirical and theoretical.

Don't forget Deepbit PPS has a 10% fee, so expect to get 10% less than bitcoins.lc over a long time period.

There are hot and cold tables though. My last trip to Reno I got ate alive by this one dealer hand after hand. When that dealer was not at the table I would get ahead even though I play like crap. Now if you can bring that analogy back into Bitcoin, well, I doubt it.

IMO there is really no reason to jump pools. You will only lose shares in the time spent reconnecting to new pools.

Over time, we all make the exact same amount of money.

Anybody who jumps from one roulette wheel after betting on red to another roulette wheel to bet on red is just plain insane.

then again, bitminter had amazing luck and I got 2 BTC mining a couple of days with 150mhash/s because it was proportional and two blocks were found in like 200,000 shares each which is much better than the expected amount

Which was the reason for setting up Deep in the first place, some of these smaller pools are starting to take days, weeks or longer.

Congratulations, you just discovered why pool hopping can pay off...

I have been using bitcoins.lc as my primary pool for awhile but decided to see what my payout would be at Deep since the last difficulty change had bitcoins.lc taking 10 hours plus on many blocks. In doing so "technically" the payout should be the same over a period of time given that one pool having a smaller hash rate means my PPS is higher but each credit will take longer, the other having a higher (much higher) hash rate so my PPS should be lower but credit for each block will be sooner, correct?

Granted there will be some fluctuation in that very basic statement above however I should assume given my hash rate is consistent with consistent stales my overall payout would remain the same, within reason, no matter what pool I used, correct (ignoring fees and so forth)?

The reason for asking is unless I need to run a far longer test or my math is WAY off I am seeing my payout at bitcoins.lc being 2x if not a little higher over a given period on Deep. Where is my logic incorrect?

Edit: I really hate having to correct my own post within minutes of posting it. Due to Deeps delayed reporting my payout is much closer than the difference cited above. It's still better at bitcoins.lc but the difference is probably within the "luck" factor, somewhere around 1.3x better, not the 2x plus I was seeing initially. I'll leave the thread though as I am interested in comments.