[1050 TH] BitMinter.com [1% PPLNS,Pays TxFees +MergedMining,Stratum,GBT,vardiff]

[organofcorti]

You seem to be obsessed with having a pedantic debate on statistics.  I have absolutely no interest in that.  If you want to CONSTRUCTIVELY discuss how public pools can be defended in an environment where people are actively trying to cheat them I would be very interested in seeing what you think.

Bitcoin mining *is* statistics. If you don't understand them, you don't understand mining.

[1714929667]

1714929667

[1714929667]

1714929667

[1714929667]

1714929667

[1714929667]

1714929667

[Entropy-uc]

You seem to be obsessed with having a pedantic debate on statistics.  I have absolutely no interest in that.  If you want to CONSTRUCTIVELY discuss how public pools can be defended in an environment where people are actively trying to cheat them I would be very interested in seeing what you think.

Bitcoin mining *is* statistics. If you don't understand them, you don't understand mining.

As I said, pedantic.

And with a Ph.D in engineering and over a decade in semiconductors I can assure you I understand the statistics just fine.

I look forward to your analysis of what actions a pool should take to defend against exploitation by miners either intentionally withholding block solutions or operating defective hardware.

[georgem]

Thank you koi, for yet another block. 

You deserve your own meme:

[hamburgerhelper]

It would be interesting to see how many users are up 75 BTC on payouts versus blocks solved.  It most certainly won't be 5%.  I doubt it is 1 in 1000.  And anyone in that kind of extreme situation should be scrutinized.

I think this illustrates why the multipool situation was unique. The magnitude of the pool percentage was very high as was the magnitude of his bad luck. Presence or absence of malfeasance aside, this combination made a few of us uneasy.

Maybe the per-user CDF should only appear for users who meet a percentage criteria? Like users over 10%(?) of the pool should always have their CDF visible on the site. I think most of us would get used to the ebb and flow of the CDF of the the big actors who are clearly here to find blocks. For the newbies and the torch n' pitchfork crowd, the rest of us can talk them off the ledge.

I really don't want to have a witch-o-meter on the site, but I also can't afford to be robbed blind. I am in favor of a rational discussion of possible solutions here. I think my idea has merit and I'd like the Doc to consider it or some variation.

[philipma1957]

I had proposed that Doc make per user CDF available on the Bitminer site. I don't think this is the whole answer, but I think it would be a start to building a systematic approach to the block withholding threat. The good doctor can explain why he didn't like the idea if he's so inclined.

So, anyone who experiences negative variance will be accused of being witches block withholders? I see that working well. Given recent history, about 8% of the pool would be so accused. If you keep kicking off the lowest performing 8% of the pool it wont be long until you have just one miner left.

Look I am hashing here with 420gh  so we are talking  a long time for a block  not even weeks or months more like a decade.

 Only  a few 100 or so have more then 1th.      1th takes almost  2 years with average luck to make a block.  

 So good cdf testing only works for those with 97 th  

 normal luck   is  1 block a week for 97 th.   8 weeks of no blocks makes one suspect if they are hashing at 97th.  

  for my 420gh 16 years .  So cdf does not work.  Since  most miners here too small to measure well.

  So  right now if we did  post cdf results  the number 2 miner should make a block every 17 days since he is at  39th.

 A fair trial for him is 8x that or 136 days.

    Like it or not  auto-switching to a solo section  of a mining pool is going to be the only way to sift out  the weeds.

We  could be in the solo section of bitminter 10% of the time and in the merged part of bitminter 90% of the time.    

yeah my 420gh will most likely never   hit while I am in the solo part  but that would discourage  someone directing his defective gear/software here.

 I can't figure any other   fair leveling method that would only attack  bad gear/bad software.      

I really would like to see a dozen top pools doing this.  I know the fear is it chases away people

big miners may say fuck it if I  want to solo mine I don't need a pool forcing me to do it I can do it myself.

WELL if big miners think a pool is filled with bad gear/ bad software they will end up doing solo mining anyway.


 OTHER SUGGESTIONS ARE WELCOME.  

Say 15% solo 85 % solo    instead of 10/90   or maybe 20/80.


 Remember when you are in the solo section If your gear is 2 watts or better per hash it is like getting more then 1100 to one one a 3 digit number bet. when in fact the normal odds are 1000 to one.

  A better idea is welcome , but I don't really see one that could be applied across most of the network.

[Bigsky]

I personally am glad to see "multipool" go.
 

Why?

Not just for the fact that he only found one block in nearly two weeks
at 200Ths, which is probabilty in itself of around 5%

Ok I will accept that as bad luck.

Questions were raised here after Multipool had not found a block for
around 5 days, then multipool found a block a day or so later to
easy everyones minds, and didnt find another block to this day.

After he found that 'sole' block, bitminter pool found another block 7mins later,
that block turned out to orphaned, we have only had two orphaned blocks
this year, the other orphaned block was not long after Multipool joined
our pool two weeks ago.

More bad luck, perhaps

Multipool in itself mines multiple coins, but seemed to have most of their power
if not all of their power pointed at bitminter for long periods of time, it appears
he pays bitminter fees, as well as the extra fees for orphaned blocks and to get paid striaght away.

Why would anyone with this much hash power do this, it does not make sense?.

A very large percentage of multipools own users use just 4 or 5 digit user names, I have
never seen this before, it looks sus to me.

The 30 block payout figure for multipool is around .06btc per Th per day, this is pretty good
figure considering how unlucky we have been, and most of his power looks to have been
pointed our way, does he somehow merge mine other coins as well.

The above just raises to many red flags for mine.

We dont know the software he uses to switch between pools, that may have issues that
affect the way our pool mines, he may not know if there is an actual problem

The four and five digit user names at multipool may have bought the same hardware
or be somehow related, I dont have a clue, but it looks very strange


But for mine at a distance it looks like a duck,
it seems to walks like a duck,

have not heard it quack yet

but it is either a duck

or I am a goose.


cheers


If I have made any errors in the above information please post them and I will issue a correction






 
  

  

[philipma1957]

A variation on the solo idea is this every day has 24 hours.

 pick  2 hours a day let it roll forward  

 0:00-  2:00   day 1
 2:00-  4:00   day 2
 4:00-  6:00   day 3
 6:00-  8:00   day 4
 8:00-  10:00 day 5
10:00- 12:00  day 6
12:00- 14:00  day 7
14:00- 16:00  day 8
16:00- 18:00  day 9
18:00 - 20:00 day 10
20:00- 22:00  day 11
22:00-24:00  day 12


in that 2 hour period whomever hits the block or blocks     it is his  or hers just as if  it was  solo mined.

yeah koi will get the most hits   he has 200 th     I have .4 th  but any one with dud gear/dud software will never get a block in that time period.


 easier to do then a second server for a second pool…   it would certainly discourage bad gear/software miners that know their gear is bad.


How about this idea Doc.  does it have merit.


   you make the same money and pay out the same money.  No extra gear or paperwork.


     2 hours a day we all get a shot at a big hit.

 the other 22 hours we are at normal.

This would discourage  that suspect bad luck canon would it not.  Opinions?

[14er]

Been lurking awhile... but the whole multipool saga provokes me to speak up.

A constructive discussion on how the pool maintains integrity is needed.  Bad or incompetent actors are out there, and we need to have a way of screening for them.  To not have a process or method in place is a disservice to the honest actors on this pool.

Pool operation rests on a couple of key premises, i think:

• 1.  all hardware is capable of finding a block or doing work to confirm a block
• 2.  all users contribute their finds to the pool
• 3.  the pool fairly tracks the hash contribution on an appropriate time basis

Are there more operating assumptions?

First, how do we validate the first two premises?  To me, over a given period of time, for a given integrated hash contribution, you should be able to calculate the expected block finds to a reasonable standard (a confidence interval or t-test).

To Entropy's point, a 95% confidence interval is a pretty well-established standard for making significant decisions.  What is the board's expectation.

I think we all want to have confidence our hardware is operating in a statistically normal way.  Bad hardware is bad for all of us, just like a bad member is bad for all of us.

So, second, how do encourage new contributors with confidence that we are not being screwed?

One idea is to first group our hashers into subgroups of 200 +/- 10% Th (terahash).  We could then determine, over a period of time, whether any subgroup is statistically deviant.  You could even randomize the groups to try and isolate weak (statistically improbable) hashers.

For newer members, we might want to consider putting them in a probationary phase (as part of a newbie group), with their payouts going to escrow until the meet a certain confidence threshold.

Ideally, we should be able to just straight-up test the member hardware with test cases, especially from the higher difficulty regime.

I think Bitminter is a solid group.  As far as I can tell, I think the contributions have been largely proportional to the member hashrates, and our hash size relative to the global community.  We may in fact be doing a little better than average recently (excluding multipool)... which begs the question - why?

[timmah]

Humm, even my 27-28th has mined at least 2 blocks that i know of... this is also only been mining for a bit over 3 weeks.

Total BTC received from the pool is just a little over 39 btc.  I know things are suppose to even out and work in our favor eventually but that just makes me think, I could have over 50 btc (if not more) if I took the risk and solo mined.

With that said Multipool has been mining for about the same time with almost 10x my hash power so how can this be right?

In any case is there some post on here detailing exactly how valid hashes can be submitted but never generate a block... I know that people have said it was software or hardware but I don't really get how someone can submit and the pool accepting hash that will never generate a block?  I thought for the most part CgMiner is used for many asics, I would doubt that he would write bad code.. also doesn't a valid hash have to have the correct amount of leading zeros (if that even matters)?  AFIK.

So it would be great if someone point me to how this is done, as far as i know the asics don't care about anything else but generate valid hashes (some of which should generate a block).  I think some very detailed super geek technical information on how this can happen would help us all especially to potentially figure out who is shooting blanks!

[Phedwell]

in that 2 hour period whomever hits the block or blocks     it is his  or hers just as if  it was  solo mined.

yeah koi will get the most hits   he has 200 th     I have .4 th  but any one with dud gear/dud software will never get a block in that time period.
<snip>
     2 hours a day we all get a shot at a big hit.

 the other 22 hours we are at normal.

This would discourage  that suspect bad luck canon would it not.  Opinions?

I would not mine here. I have .6 th. That's like an extra 8.333% donation (not that donations are bad or anything). But with so little hash I'll never find a block. I'm here for the 'pool' aspect of the pool.

It's like adding the Perk, 'Feeling Lucky'
Donations required - 8.333%.

I'm not feeling that lucky. But perhaps if all the people that worked a two hour shift were to pool 'their' blocks...

[georgem]

holy moly kumbaya... 4 blocks within 12 hours? 

[Minor Miner]

holy moly kumbaya... 4 blocks within 12 hours? 

Yes, I feel terrible.    Karma.   And we are on the good side.
I like how a lot of new accounts are commenting about multi pool.   Either it is stirring a lot of interest or people who are semi known do not want their thoughts associated with themselves.
OOC, we are not completely stupid.   The situation is unique to large miners.    We believe there is bad software (proven) written by many in the rush to market (and that it is not all found) and we also believe there are some bad mining equipment that cannot be fixed.   I think you know how we are so sure of the two above "assumptions" and why we are so sure they are correct.   
People who have large mines are the most obvious (like the large groups at BTCGuild and Eligius), they are the easiest to find.   BUT, there needs to be a confirmation process (which the current pool software does NOT allow for) because it is highly likely in the future someone will unknowingly sell miners that just do not work and they will be spread to users that do not know.
This will dilute everyone and will be a large problem as the difficulty curve approaches the point where marginal returns on electric are slim.

I proposed to solve the BTCGuild problem of months ago that all the largest unluckiest miners were sent a test (a previously solved block) to see if they all could potentially solve it.    I was told that this was not possible (in my mind a flaw in the software).    Some other solution needs to be found.    We are getting close to the point where 100TH/s will cost people $10-15,000 a month in electricity and will only solve a block a month (likely 3-4 months away from that).   Solomining is extremely risky when we get to that point as variance requires a lot of working capital.   Pools are essential but I think need to be fundamentally changed.

[AaronS]

And expect to hear from me next time Koi hits a dry spell.  Although I expect you'll both be conspicuously absent from the forum if that happens.

I can take it and I will still be here, but hopefully it does not happen until we are 2% of the pool. 
Let me know on those stats after you crunch the numbers.   I will be interested to know (maybe you have some reason for me to "hear from you" now).    As you have said, I am a moron that cannot do simple math so help me out here.
Hopefully, everyone will become the neighborhood watchers, so I would encourage you to call out anomalies and let OOC tell you that it is not likely (although his 8X is more than anyone can really stomach from someone over 10% of a pool).   Hopefully, you do not go 15 more days to get to the point where OOC would agree "something is likely strange..."

I don't have any problem with Koi's numbers.  He appears to be doing rather well lately.  That said, if he's so confident in his ability to find blocks at a rate above statistical average over long periods of time, one might wonder why he doesn't just skip pooled mining altogether and solo mine.

Now you are truly being a fucking idiot, stop while only a few people think you the fool.   Show the quote where I said that.  What idiot would believe that?
 I have absolute confidence that over the long run, KOI will solve the number of blocks consistent with our hash.   Because we have tested the equipment we use and the software we use.   Multipool?  You cannot make that same assertion because you DO NOT KNOW.

I did not think you could do the stats you demanded to do on Koi.   Or are you too lazy?   Like I said, put up or STFU.
Why don't you give us the math on your claim that we all would make less if you left the pool?   That should be interesting to learn for us all.

You didn't say it, but Entropy-uc did:

Your numbers say that 92% of the time a honest contributor would not have performed so poorly.  That's is enough reason to freeze payments until more information is available.

If DrH was to freeze funds for every miner that ever had reached the point where they had submitted 2.5*D shares and only solved one block, nearly everyone who has solved more than one block would be very pissed off.

I can almost guarantee you that at some point you will have submitted 2.5D shares for one block.

I can guarantee you we have solved blocks worth substantially more than we have been paid.

Multipool has been paid 4x what they have contributed to the pool.  Maybe you think a $50k donation from all the other users of this pool is trivial.  I do not.

[Puppet]

I mentioned before that, in the days of 50 btc per block rewards, I solved one block, but did 3 blocks worth of work. On most pools I solved no blocks at all. Did I cheat those pools out of $50 000? Are you also 95% certain that the 5% of pool members that have had similarly bad luck are also cheating this pool?

Its clear to me this problem cant be solved with statistics. Variance of good faith small hashrates is far too great, and anyone with bad faith and large hashrate can easily divide his hashpower over as many accounts/workers as he wants.

I dont know how stratum works at all, but the only solution I can think off that would be 100% watertight is changing the protocol in such a way that the miner can not know if his share solves a block or a not, and only the pool can determine that. Is that something thats even theoretically possible to implement?

[eleuthria]

I dont know how stratum works at all, but the only solution I can think off that would be 100% watertight is changing the protocol in such a way that the miner can not know if his share solves a block or a not, and only the pool can determine that. Is that something thats even theoretically possible to implement?

Not without a Bitcoin hardfork.  Miners inherently have to know if their hash solves a block, because they need to know if their hash meets the difficulty the pool expects.  It's the same string of binary either way, one just has more 0-bits in front than the other.

[Minor Miner]

And expect to hear from me next time Koi hits a dry spell.  Although I expect you'll both be conspicuously absent from the forum if that happens.

I can take it and I will still be here, but hopefully it does not happen until we are 2% of the pool. 
Let me know on those stats after you crunch the numbers.   I will be interested to know (maybe you have some reason for me to "hear from you" now).    As you have said, I am a moron that cannot do simple math so help me out here.
Hopefully, everyone will become the neighborhood watchers, so I would encourage you to call out anomalies and let OOC tell you that it is not likely (although his 8X is more than anyone can really stomach from someone over 10% of a pool).   Hopefully, you do not go 15 more days to get to the point where OOC would agree "something is likely strange..."

I don't have any problem with Koi's numbers.  He appears to be doing rather well lately.  That said, if he's so confident in his ability to find blocks at a rate above statistical average over long periods of time, one might wonder why he doesn't just skip pooled mining altogether and solo mine.

Now you are truly being a fucking idiot, stop while only a few people think you the fool.   Show the quote where I said that.  What idiot would believe that?
 I have absolute confidence that over the long run, KOI will solve the number of blocks consistent with our hash.   Because we have tested the equipment we use and the software we use.   Multipool?  You cannot make that same assertion because you DO NOT KNOW.

I did not think you could do the stats you demanded to do on Koi.   Or are you too lazy?   Like I said, put up or STFU.
Why don't you give us the math on your claim that we all would make less if you left the pool?   That should be interesting to learn for us all.

You didn't say it, but Entropy-uc did:

Your numbers say that 92% of the time a honest contributor would not have performed so poorly.  That's is enough reason to freeze payments until more information is available.

If DrH was to freeze funds for every miner that ever had reached the point where they had submitted 2.5*D shares and only solved one block, nearly everyone who has solved more than one block would be very pissed off.

I can almost guarantee you that at some point you will have submitted 2.5D shares for one block.

I can guarantee you we have solved blocks worth substantially more than we have been paid.

Multipool has been paid 4x what they have contributed to the pool.  Maybe you think a $50k donation from all the other users of this pool is trivial.  I do not.

Is this still you Adam or another fool?    Can you read?    He said solved.   That is an analysis of past results.  This is called a fact.   Not a prediction.    You need to learn to read and comprehend.   

[philipma1957]

in that 2 hour period whomever hits the block or blocks     it is his  or hers just as if  it was  solo mined.

yeah koi will get the most hits   he has 200 th     I have .4 th  but any one with dud gear/dud software will never get a block in that time period.
<snip>
     2 hours a day we all get a shot at a big hit.

 the other 22 hours we are at normal.

This would discourage  that suspect bad luck canon would it not.  Opinions?

I would not mine here. I have .6 th. That's like an extra 8.333% donation (not that donations are bad or anything). But with so little hash I'll never find a block. I'm here for the 'pool' aspect of the pool.

It's like adding the Perk, 'Feeling Lucky'
Donations required - 8.333%.

I'm not feeling that lucky. But perhaps if all the people that worked a two hour shift were to pool 'their' blocks...

  This why I would prefer the top dozen pools to do it.  Thus you would not be able to avoid it or you  would have to go to a really small pool with high variance anyway.

 

I also want you to realize that during the 3 hour solo pool you are getting  better odds  then the  power spent by your gear.

I am not going to get back in the top 50 here and I was in the top fifty from sept 2012 to may 2013  using a lot of gpus  20 -25 and 11-13 pc's.

  Mining coins as low as 6 bucks a piece. I paid a lot of dues for BTC. I taught a few hundred people how to mine once I sold them gear on ebay.

Here is my ebay record.

http://feedback.ebay.com/ws/eBayISAPI.dll?ViewFeedback2&userid=philipma1957

Here is my bitcointalk trust  rating

https://bitcointalk.org/index.php?action=trust;u=64507


  I support BTC for more then just the  "I will make money idea of it".




Back to ideas splitting us into 5 teams with 20 percent of the hash  each can help.  

 But  I would like to propose another idea.  I have 3 ant miner s-1's running under clocked and undervolted. 420gh how about  we get a lot of antminer s-1 guys group them and see how ant miners do.

I have  6 r-box  they run 195gh  I can run them here in an r-box section along with a lot of other r-box

I have 1 dragon miner hosted 1th I run it at btc guild I am will to point it at a dragon miner only section of a pool.
 

  This idea is more work to do you need a miner to show some proof of what gear he really has. The sections need to be large 100th or more. since right now 100th is about a block a week.  you need to run around 8 weeks to see if a 100th group is really far off.

   This idea also has some other merits if you find a trouble section  you can further test for  possible  commonalities within the group.  
Lets say hard wired vs wi fi.  may turn out that 100 sp10's with hard wired connections work better then 100 sp10's with wi fi.
On in the case of my hosted dragon miner it could turn out that it can't make a block. while people using dragon miners in house make blocks 2x the rate.

Once again it is harder to do then the simple 1/12 of your work  is  done solo mining idea that  I offered earlier.

[CYPER]

Just out of curiosity, what is the latest drama you are all discussing? 50K donation? What happened?

[DrHaribo]

With that said Multipool has been mining for about the same time with almost 10x my hash power so how can this be right?

If you and a friend walk into a casino, your friend bets 100 USD and wins 1 USD, you bet 1 USD and win 100 USD. How can this be right?

I don't think the casino would even bother to explain it. They would look at you for a moment, and decide you have had enough to drink.

It's random. This is why you mine in a pool.

Some users get a lot more pay than the blocks they find, some get a lot less. This is how pools work.

95% CDF doesn't mean there is 95% probability that the person is a scammer. If you mine long enough, it will happen to you. It doesn't mean you are a scammer. 5% of the time your luck is expected to be that bad.

[14er]

On Phillip's idea to assess miners by the hardware they use:

As a HW consumer and pool user, I would love that comparison, but I don't see how that can be done in practice in anonymous pools.  Moreover, the performance capability of miner is affected, as Minor_Miner points out, to hardware and software, as well as clocking conditions.

I just don't see a way of validating that users are grouping their hardware correctly.

The simplest / cleanest way of validating a user's hardware is to be able to periodically and randomly test a user's hardware / hash volume with a test case (like drug testing for athletes). A larger user should have to demo hashes proportional to their size.

The next level (but more complex) option is to group users into larger buckets (5 Th sub groups and 250 Th groups) to assure one bucket is behaving in statistically consistent ways, with randomization of the subgroups periodically to ferret out weaker players.  This idea will go to address the viability of the hardware as we climb the difficulty curve.  It is conceivable that hardware that works at lower difficulties breaks at upper difficulties simply by the way the HW/SW combo hashes.  We are looking for unexpected / forseen issues with hardware.

I expect the latter problem to go away as the payouts diminish with time, and the power hungry machines start to lose money on an operational (electricity cost) basis.  The more mature stuff *should* be better tested.