Eligius: 0% Fee BTC, 105% PPS NMC, No registration, CPPSRB

[praeluceo]

Thank you for the quick update Luke! I know that luck isn't the same as stale blocks, but we did have 3 stale blocks this week, which feels higher than average at least, so I thought it'd be good to compare to Slush's update today.

So Slush's Pool posted a news update regarding recent orphaned blocks:

https://mining.bitcoin.cz/news/recent-orphaned-blocks-explanation/

I'd be very interested to hear from wizkid057 or Luke-Jr on whether we are also watching our pool; what version of bitcoind Eligius is based off, and whether block submission propagation times are being monitored at all. I'm not just asking because of our terrible luck as of late, which I understand is entirely by chance, but also because having more information on the internal operation of the pool and block submission/fast block relay statistics!

The main bitcoind is running a 0.10.1-based bitcoind. (I don't know how to get info on the other bitcoinds.)

Note that luck is not related to stale blocks (what slush erroneously calls orphaned/invalid).

[1714010632]

1714010632

[1714010632]

1714010632

[1714010632]

1714010632

[1714010632]

1714010632

[Biodom]

Well, how about that. It should have been 1/1000 weeks chance or 1/20 years roughly according to ogranofcorti if i understand him correctly.
If I was running a pool, I would double check everything and updated to whatever is working for others, if necessary.

[Grix]

http://organofcorti.blogspot.no/2015/07/faq-bitcoin-mining-and-luck-statistic.html

So I read this and as far I understood calculated the approximate chance of being as unlucky as eligius have been:

http://www.wolframalpha.com/input/?i=CDF(ErlangDistribution(2000%2C+2000)%2C+1.0487)

Apparently it's 1.5%..
65 in every 66 times the pool would be more lucky than it has been last year. Confirms the feeling that it has been awful..

Fun fact, I did the calculation again and now it's 0.7%

http://www.wolframalpha.com/input/?i=CDF%28ErlangDistribution%282000%2C+2000%29%2C+1.0559

That's right, in only 10 days the luck has been so bad that it straight up halved the percentile built up over a whole year of mining.

I know posting this won't make any difference, but I feel like we've earned some whining rights by now.

[RealMalatesta]

I know posting this won't make any difference, but I feel like we've earned some whining rights by now.

And with changing into winter time here in Europe, we've got one more hour for whining...

[wizkid057]

As much as I wish there was something that could be done about bad luck, there really is nothing to be done.  If the pool op could control if the pool was lucky or not... well, that'd be great... but would probably defeat the purpose of mining if it were possible.

I've been checking and double checking everything pool side as I've been making improvements to the back end.  The pool is almost entirely on new servers now vs a month ago.  Slightly fatter bandwidth pipes and all to better handle the huge bursts of outbound bandwidth that accompanies updating work on over 100,000 workers connected to the pool.

I investigate every stale block the pool finds.  I've not seen anything out of the ordinary. Just losing some races with the Chinese pools a couple of times where they confirmed their own blocks.  Unfortunately, normal.

I have a suspicion that some mining hardware out there may be software time bombed and have miners who are using the hardware unknowingly withholding blocks sometimes.  Unfortunately there is next to zero chance of verifying this in our lifetimes.  Given that luck has been crap across basically every pool with accessible stats I think it's a pretty plausible conspiracy theory. It also makes some logical sense given that some of the largest miners today are sellers of ASIC mining hardware.  Basically would effectively cut into their customers' profits (withholding some of their blocks) and increase their own (by keeping the network difficulty increase lower than the amount of hardware sold, plus they could obviously disable any time-bomb code when using the hardware for themselves).

An individual entity performing a withholding attack probably wouldn't cause such widespread poor luck, and I don't see any conclusive evidence of any individual withholders on Eligius using the algorithm's I wrote for investigating potential withholding (which use much more data that just the username).

Anyway.  I caught up BTC payouts the other day.  Caught up NMC payouts today.  And things should be business as usual around here.  Would be great to find more hash power, but with the super low margins on mining these days I'm not all that hopeful, unfortunately.

[Biodom]

I have a suspicion that some mining hardware out there may be software time bombed and have miners who are using the hardware unknowingly withholding blocks sometimes.  Unfortunately there is next to zero chance of verifying this in our lifetimes.  Given that luck has been crap across basically every pool with accessible stats I think it's a pretty plausible conspiracy theory. It also makes some logical sense given that some of the largest miners today are sellers of ASIC mining hardware.  Basically would effectively cut into their customers' profits (withholding some of their blocks) and increase their own (by keeping the network difficulty increase lower than the amount of hardware sold, plus they could obviously disable any time-bomb code when using the hardware for themselves).

interesting allegation...it would be absolutely scandalous if true, to the point of no one in their sane mind would buy anything from that company again.
Can any sophisticated coder explore the binaries to check for such potentially withholding code?

[Luke-Jr]

I have a suspicion that some mining hardware out there may be software time bombed and have miners who are using the hardware unknowingly withholding blocks sometimes.  Unfortunately there is next to zero chance of verifying this in our lifetimes.  Given that luck has been crap across basically every pool with accessible stats I think it's a pretty plausible conspiracy theory. It also makes some logical sense given that some of the largest miners today are sellers of ASIC mining hardware.  Basically would effectively cut into their customers' profits (withholding some of their blocks) and increase their own (by keeping the network difficulty increase lower than the amount of hardware sold, plus they could obviously disable any time-bomb code when using the hardware for themselves).

interesting allegation...it would be absolutely scandalous if true, to the point of no one in their sane mind would buy anything from that company again.
Can any sophisticated coder explore the binaries to check for such potentially withholding code?

Which standalone miners are we talking about?

Spondoolies lacks motive - afaik they aren't self-mining. Also their minergate has been open source from the start.

Bitmain recently released complete source code for at least the S5 controller.
Their previously-closed kernel module is aware of blocks, but only writes to a log when it sees them - nothing seemed suspicious there.
It's conceivably possible that it used to do something different before the code was published, but...
Their self-mining operation is so large, they would need to have their admins install non-public firmware images, and someone I know who has such a position was never told not to use the public firmware.

BFL's Monarchs are controlled by a PC, and the miner itself has no way to really do this.

Avalon 2 and newer do most of the heavy lifting in FPGA, but that is open source and I didn't notice anything odd looking at it a while ago.

[MCHouston]

Wow this pool has lost over 6 PH in less then a week. Its a shame really. Its a decent pool.

[RealMalatesta]

I have a suspicion that some mining hardware out there may be software time bombed and have miners who are using the hardware unknowingly withholding blocks sometimes.  Unfortunately there is next to zero chance of verifying this in our lifetimes.  Given that luck has been crap across basically every pool with accessible stats I think it's a pretty plausible conspiracy theory. It also makes some logical sense given that some of the largest miners today are sellers of ASIC mining hardware.  Basically would effectively cut into their customers' profits (withholding some of their blocks) and increase their own (by keeping the network difficulty increase lower than the amount of hardware sold, plus they could obviously disable any time-bomb code when using the hardware for themselves).

interesting allegation...it would be absolutely scandalous if true, to the point of no one in their sane mind would buy anything from that company again.
Can any sophisticated coder explore the binaries to check for such potentially withholding code?

Which standalone miners are we talking about?

Spondoolies lacks motive - afaik they aren't self-mining. Also their minergate has been open source from the start.

Bitmain recently released complete source code for at least the S5 controller.
Their previously-closed kernel module is aware of blocks, but only writes to a log when it sees them - nothing seemed suspicious there.
It's conceivably possible that it used to do something different before the code was published, but...
Their self-mining operation is so large, they would need to have their admins install non-public firmware images, and someone I know who has such a position was never told not to use the public firmware.

BFL's Monarchs are controlled by a PC, and the miner itself has no way to really do this.

Avalon 2 and newer do most of the heavy lifting in FPGA, but that is open source and I didn't notice anything odd looking at it a while ago.

I discussed the same suspicion a while ago. Maybe a silly question, but: Wouldn't it be possible to put some code IN a chip - be it an ASIC or a chip with another task - which could be triggered if a certain call is broadcasted?

An alternative scenario would be that there is a weakness in some controller software of which most programmers aren't aware. Don't say it is not possible - ssh taught us something, didn't it? If someone knows about this weakness, he could use it.

However, as long as producers of miners also run their own pools, this allegation will always come up.

[wizkid057]

I have a suspicion that some mining hardware out there may be software time bombed and have miners who are using the hardware unknowingly withholding blocks sometimes.  Unfortunately there is next to zero chance of verifying this in our lifetimes.  Given that luck has been crap across basically every pool with accessible stats I think it's a pretty plausible conspiracy theory. It also makes some logical sense given that some of the largest miners today are sellers of ASIC mining hardware.  Basically would effectively cut into their customers' profits (withholding some of their blocks) and increase their own (by keeping the network difficulty increase lower than the amount of hardware sold, plus they could obviously disable any time-bomb code when using the hardware for themselves).

interesting allegation...it would be absolutely scandalous if true, to the point of no one in their sane mind would buy anything from that company again.
Can any sophisticated coder explore the binaries to check for such potentially withholding code?

Which standalone miners are we talking about?

Spondoolies lacks motive - afaik they aren't self-mining. Also their minergate has been open source from the start.

Bitmain recently released complete source code for at least the S5 controller.
Their previously-closed kernel module is aware of blocks, but only writes to a log when it sees them - nothing seemed suspicious there.
It's conceivably possible that it used to do something different before the code was published, but...
Their self-mining operation is so large, they would need to have their admins install non-public firmware images, and someone I know who has such a position was never told not to use the public firmware.

BFL's Monarchs are controlled by a PC, and the miner itself has no way to really do this.

Avalon 2 and newer do most of the heavy lifting in FPGA, but that is open source and I didn't notice anything odd looking at it a while ago.

I probably should have made clear that the existence of an inadvertent time bomb is possible, as well as a nefarious one.

As for alternate firmware for the nefarious purpose, I'd figure that it would just be some command/flag/whatever set that would disable the time bomb.

[not.you]

I don't know enough about the technical aspects to say if this would work or not but I thought of a simple way for a mining manufacturer to sabotage miners.  If the software in the miner told the miner to multiply the network difficulty by a factor, say 1.2 it would have the effect of making the miner think some valid blocks were not actually valid.  But I am under the impression these shares would still be submitted as valid shares since they would exceed whatever the miner's share difficulty was.  So the part I don't know, does the miner flag a block in some way as different than a regular share when it sends it to the pool?  So if all of these miners were basically working from the assumption that network difficulty was much higher than it actually was, they would still find blocks but not nearly as many.  And this should be visible on the pool end if a large miner who seems to have bad luck but not impossible luck only submits blocks that are well higher then the actual network difficulty.  Also if you were the manufacturer of this type of equipment you could just make sure your pool checked high difficulty shares that were close but not sent as blocks.  Or alternatively maybe just tell your workers that the network difficulty was lower to offset the multiplying factor built into the equipment.  In that case you could still have hardware from other manufacturers on the same pool as long as you made sure to validate the blocks the workers submitted since some of them would then be lower than the actual network difficulty, and validating blocks I would assume would already be something a pool is doing anyway.

Just my technically uneducated musings on the topic...

[btcmner]

I don't know enough about the technical aspects to say if this would work or not but I thought of a simple way for a mining manufacturer to sabotage miners.  If the software in the miner told the miner to multiply the network difficulty by a factor, say 1.2 it would have the effect of making the miner think some valid blocks were not actually valid. 

That's an interesting idea, and it is also easy to test, by checking the difficulty of the blocks found lately. Well, that seems pretty normal to me, so this hypothesis can safely be pushed under the rug.

[notabeliever]

I don't know enough about the technical aspects to say if this would work or not but I thought of a simple way for a mining manufacturer to sabotage miners.  If the software in the miner told the miner to multiply the network difficulty by a factor, say 1.2 it would have the effect of making the miner think some valid blocks were not actually valid.  

That's an interesting idea, and it is also easy to test, by checking the difficulty of the blocks found lately. Well, that seems pretty normal to me, so this hypothesis can safely be pushed under the rug.

how about the affect of the BitcoinXT Fork

or someone needs to keep the price of BTC below 300

[wizkid057]

Definitely seems to be some confusion lately on how mining works in general.  Seems to be a large portion of people buying mining hardware that have absolutely no clue what they're doing, no clue how mining even works, sometimes no clue about bitcoin in general, etc.

I got an email from someone this morning insisting that Eligius cheated him out of something like 50 BTC.  Aside from the usual "lol" I get from these emails, it's getting a little irritating because they're becoming more frequent.  Details in this case being that he bought mining hardware a couple of weeks ago at a ridiculously over priced cost, less than 100 Gh/sec worth that will *never* break even, and was expecting returns of something like 10 BTC/day ("over $2k per day" according to his email).  Whoever is peddling this mining hardware to people who don't know any better with the get rich quick smile on their faces should be ashamed of themselves.  All too often I end up being the person that has to tell these folks that they took it up the back side when they bought their hardware, and usually these people are not grateful and still try to believe that somehow I'm the one screwing them over.... at which point I just ignore further communication.  Gets pretty old.

Also a lot of confusion from people in the middle who somewhat understand how mining works..... about how mining works.

I don't think people understand that there are no longer any straight PPS mining pools in existence anymore, for obvious reasons.  None that I'm aware of anyway, which means none large enough to not be a scam probably.  Yet at least once per day I get an email/support ticket/whatever from someone saying something like, "I have X Th/sec of hardware, I should be making Y BTC/day."  And usually this is after something like 2 days or less of mining.  Seems many people mining today have no idea about variance.  There is no way to earn a set amount per day with a set amount of hash power.  It just isn't going to happen.

Pools iron out variance to an extent.  Eligius's CPPSRB reward system, even with today's high network difficulties and relatively low percentage of the network, still comes out ahead in all of my simulations.  It's the most consistent you can be, outside of straight pool-takes-all-risk PPS.  But I'll still get someone mining for less than one day (sometimes just a couple of hours!) sending me messages about how they haven't been paid exactly what they should have been.  Someone even emailed me a while back after a month insisting the pool was broken because based on his hardware's manufacturer advertised hash rate he should have earned 0.7xxxxxxx BTC and he actually was only paid 0.698xxxxx BTC (shares rewarded were actually 99.9%).  Surprisingly his math was fine, and his hash rate was nearly what he was supposed to be getting (not quite 100%, but close)... and just didn't understand how things worked well enough to understand that his earnings were completely normal.  He earned 99.9% of his expected earnings, but this wasn't good enough.  He switched to a pool with a fee.  I just shook my head and laughed.

Anyway.  Just a mild rant from me.  I guess these miner education issues aren't really going to change any time soon, especially with smaller miners being edged out.

Oh, and about 9 months left until the next block subsidy halving.  Just going to throw that out there for people buying mining hardware with 1+ year expected ROI. lol.

[RealMalatesta]

wizkid: You could forward their address to me - maybe they would buy my old fpga-towers :-p

I guess things will become worse when more gadgets with integrated ASICs will be sold. Just imagine the wifi-toaster with two asics, always mining when you make your toast... "Eligius is ripping me off! First I paid 2000+ for a toaster, ate 9230 lb toast, am obese now and didn't get a single satoshi..."

[wizkid057]

wizkid: You could forward their address to me - maybe they would buy my old fpga-towers :-p

I guess things will become worse when more gadgets with integrated ASICs will be sold. Just imagine the wifi-toaster with two asics, always mining when you make your toast... "Eligius is ripping me off! First I paid 2000+ for a toaster, ate 9230 lb toast, am obese now and didn't get a single satoshi..."

lol.  Yeah the integrated mining in devices aspect is kind of funny.  I mean, what are the odds of finding a block with 5 Gh?  Funny enough I tried asking my IRC bot and it rounded to zero. lol.  If I did the early morning math right, it comes out to, at the current difficulty, about a 1 in 20,000 chance per month.  So, probably better off just buying some scratch offs or something.

Edit: I figure the manufacturer of such devices will just pocket the reward anyway.  All things considered I guess it's not a bad plan if you're a toaster manufacturer.  100,000 toasters and you've got 125 BTC/mo coming in on average.

[sorry2xs]

I have ? at what point does the pool  determine that is lucky and begin to pay  shelved shares, i mined long enough to earn 12.07791840 BTC and that was 95.64 % as of last block earnings, i look at % as  a fee since i have not and it looks like am not going to get the 100 % mark 

[wizkid057]

I have ? at what point does the pool  determine that is lucky and begin to pay  shelved shares, i mined long enough to earn 12.07791840 BTC and that was 95.64 % as of last block earnings, i look at % as  a fee since i have not and it looks like am not going to get the 100 % mark  

And here lies another problem/misconception.  Let's say you were mining on another pool that had say a 2% fee.  So variance being what it is and landing you at 95.64% you earned 12.07791840 BTC.  Then the pool would take 2% of that, the post-variance earnings, for a fee and you would end up with 11.83636003 BTC in your pocket instead, which would be more like 93.72% share value rewarded after the fee.  You have to keep in mind that variance is NOT a fee.  It is normal and present no matter where you mine, Eligius or any other pool.  Eligius does not get the difference between 100% and your shares rewarded as a fee.  That difference is the amount of coins never actually mined by the pool due to variance and stale blocks.  If it were actually a fee... I think I'd have retired by now. lol.

As for when shelved shares are paid, any time a block is found where "luck" is > 100% shelved shares are paid.  Any time a block is found where "luck" is < 100% shelved shares are left in the share log.  Over time, variance will tend towards 100% shares rewarded.  Long term, due to stale blocks (also normal) it's impossible to maintain 100% shares rewarded (again, at ANY pool) and will usually be in the upper 90s% long term.

I've been tempted to remove the whole shares rewarded aspect from the stats since a lot of people just don't seem to understand the actual meaning behind it.  But I feel like that would be more misleading, like most pools that don't actually provide this data at all.

[Questor]

Variance being what it is, up front HW cost and ROI have to be factored into any decision to mine for more than just fun.

We took a HUGE hit on the 50% week, having launched a significant amount of hash against the pool, that will likely never be repaid.

Farms we set up earlier this year against Eligius are showing only ~80% of shares paid...  It doesn't help ROI if luck / variance takes YEARS to even out.
Sadly, math shows that running these farms against nicehash would have paid more over 3, 6, and 9 months than it has with Eligius.

LUCK is a fickle mistress, and has not been on St. Eligius' side lately.

[sorry2xs]

It has become obvious to me the big pools are here stay and an old timer told me a long time ago "if you can't beat them join them"