btc guild unlucky of late?

[stryker]

variance!! do I win a prize? 

Ah... regarding this hearsay about a large hasher withholding blocks, I was not aware of that but as I said at the beginning of this thread I was starting to wonder if it was something like that going on.

Don't get me wrong, I love the pool's site, best interface around and very clear, plus the operators at least appear to be very transparent.  It was just that terrible luck they had a few days ago that left me thinking WTF??

[1715495404]

1715495404

[1715495404]

1715495404

[1715495404]

1715495404

[DrG]

variance!! do I win a prize? 

Ah... regarding this hearsay about a large hasher withholding blocks, I was not aware of that but as I said at the beginning of this thread I was starting to wonder if it was something like that going on.

Don't get me wrong, I love the pool's site, best interface around and very clear, plus the operators at least appear to be very transparent.  It was just that terrible luck they had a few days ago that left me thinking WTF??

Well the pool was running close to 105.9% luck if I recall for their 3 month average before the sh*t hit the fan in May (when the withholding attack started supposedly).  At that time nobody complained that the pool might be cheating, although people did allege that in the past back in 2011.

So when it's up almost 7% (106-99) nobody complains.  When it's down 7% something is wrong.

The withholding attack is not hearsay BTW, the person was identified.  But he states his custom version of CGMiner was at fault.  The thing is it could have happened at any pool, and some pools don't even bother with the nice stats. Too bad the withholding didn't go on at ghash.io lol.  The same "attack" also occurred on Eligius and the guy is asking for payment which Eligius held once they discovered the problem.  You can check the pools subforum for this info.

Eleuthria might not be eligible for sainthood but he's one of the best pool ops around.

[stryker]

I've got to admit I've gone native and have been mining at ghash for a couple of days, but only out of curiosity as I've not mined there before.  One thing that hits me is everything at ghash seems opaque compared to btc guild....

For instance if you own ghash cloud hashing and also mine with your own kit sure they give you the number of hashes the ghash kit produced but its presented in such a non-intuitive way... I find myself switching between ghash and cex screens trying to marry up data...  I wrote a very simple php tallying script to produce some figures from the CSV you can download, but the CSV lacks some of the data you get on the screen.....

Once I've had my "fun" I think I'll be switching back to btc guild anyway... worth it for the figures and hassle-free mining.

[stryker]

Actually I take it back.... btc guild has only found one block today at 10:30am UK time.  Amazing, yesterday they'd found 8 blocks by now.... thats some swinging variance

[jonnybravo0311]

Actually I take it back.... btc guild has only found one block today at 10:30am UK time.  Amazing, yesterday they'd found 8 blocks by now.... thats some swinging variance

Funny how that variance stuff works, huh?

It's just the nature of the beast.  People see the words "expected earnings" and "time to block" and think, "that's what is supposed to happen."  The reality of it is that folks are far more likely to come up with conspiracy theories, than to accept reality.  Pool luck up 10%?  Somebody's cheating.  Pool luck down 10%?  Somebody's attacking.  As DrG pointed out, there indeed WAS something that happened and was discovered.  The problem now is that everyone automatically assumes that same thing is constantly happening if the pool luck so much as quivers below the expected luck.  The topic has been discussed on virtually every pool's thread, and people read it and say, "Well, that explains it then" and then start spreading the misinformation.  It's precisely why I posted that silly statement about 5 Fridays, Saturdays and Sundays.

[stryker]

As it stands BTCg has one block for today when they usually have at least 10... guess thats luck down by 90% then huh  

:edit

Actually I've just added this to the post as it looks totally sarcastic... I get the variance, I see it in larger pools like ghash... they typically have a block every 20-30 mins yet some hours they have many more and other rounds take an hour or so... guess being a bigger pool the variance plays out quicker making it easier to observe in a short time-frame.

[jonnybravo0311]

As it stands BTCg has one block for today when they usually have at least 10... guess thats luck down by 90% then huh  

:edit

Actually I've just added this to the post as it looks totally sarcastic... I get the variance, I see it in larger pools like ghash... they typically have a block every 20-30 mins yet some hours they have many more and other rounds take an hour or so... guess being a bigger pool the variance plays out quicker making it easier to observe in a short time-frame.

Being a bigger pool means the variance is far less noticeable.  Sure, GHash.io might have a block take 2 hours, but then find 4 in 20 minutes, and so on... overall, to you the miner it makes little difference because things balance out.  On smaller pools, that bad luck becomes very quickly noticed.  I mine on p2pool, where the average block time is just about a day.  (Well, right now it is 16.5 hours since the pool is at about 1.25PH/s), but we haven't found a block in just under 2 days.  It's painful.  At 16 hours a block, we should be close to finding our third... but haven't found 1 yet.  But, then again, just the other day we found 4 blocks.

[stryker]

haha I run a personal p2pool node, notice 2 have just recently been found! 

Can you give me a rough ballpark figure on what 1TH/s would get you per block?  I know it either hovers around 850TH/s or 1.2PH/s of late.  I've tried doing my sums with the data available but a 25btc payout divided by 850 yields a result which seems to high to me... looking at how many blocks p2pool solves over a typical month it looks like 1TH/s pointed there would earn almost twice what I would expect?!?

[jonnybravo0311]

haha I run a personal p2pool node, notice 2 have just recently been found! 

Can you give me a rough ballpark figure on what 1TH/s would get you per block?  I know it either hovers around 850TH/s or 1.2PH/s of late.  I've tried doing my sums with the data available but a 25btc payout divided by 850 yields a result which seems to high to me... looking at how many blocks p2pool solves over a typical month it looks like 1TH/s pointed there would earn almost twice what I would expect?!?

Hmmm... a per-block estimate.  Let's see.  First, we're going to have to make a few assumptions:

1) All shares are equally weighted
2) The pool maintains a constant 1PH/s rate
3) The share difficulty remains a constant 3.5M
4) The miner's hash rate is 1TH/s
5) The miner has been hashing for at least 3 days (so that he's got a full spread in the 8640 shares considered for payout)
6) Shares are added to the chain constantly on the p2pool network every 30 seconds
7) This miner is not the one finding blocks (i.e. he doesn't get the block finder's bonus)

Obviously as the above assumptions are changed, the results of this would also be changed.

First, let's calculate the expected time to find a share to see how many shares we can expect to have on the chain at any given time:

Code:

3500000 * s**32 / 1000000000000 = 15032.39 = seconds to find a share
259200 / 15032.39 = 17.24 = number of shares found in 3 days.  We'll just round that down to 17.

So, it would appear that we'll have 17 shares on the chain.  Now let's see the per-block reward:

Code:

25 * 199/200 = 24.875 = reward to be split amongst the non block finders
24.875 / 8640 = 0.00287905 per share
0.00287905 * 17 = 0.04894385 coins per block

So, with all of the given assumptions, you would expect about 0.04894385BTC per block.

Now, let's be perfectly clear here - my assumptions are NOT an accurate reflection of reality.  Specifically, shares are weighted, and share difficulty is constant fluctuating.  In other words, take the above with a grain of salt.  If you want to see how you're doing compared to how you should be doing, look at your node's efficiency rating.  If it's around 100%, you're good to go.  Above 100% you're doing a bit better than you should be.  Below 100% and you're doing a bit worse.

[jjc326]

I was with bitminter for a while I swear it had the worst luck for a LONG period of time.  This was about a year ago I think.  You just keep thinking it will turn around, it has to, due to variance, but it just never seemed to get a long period of good luck to make up for it.  I was happier when I left.  That is part of the reason why you get everyone going to the BIG pools, because there should be less variance.  It is a big pool problem for part time miners.

[stryker]

With your train of thought there jjc.

Thanks jonnybravo for sharing that.... I must admit my calculations are much simpler as I've not done the homework as you have but the answers were similar so its good to see that confirmed....  As I said I'm giving ghash a good weeks trial at least, I'll only cut it short if ghash approaches 50% again.  My next trial may well be p2pool, just wish that forestv fella was a bit more engaging with the community, it would be nice to know *someone* is keeping on top of the project!

[DrG]

I was with bitminter for a while I swear it had the worst luck for a LONG period of time.  This was about a year ago I think.  You just keep thinking it will turn around, it has to, due to variance, but it just never seemed to get a long period of good luck to make up for it.  I was happier when I left.  That is part of the reason why you get everyone going to the BIG pools, because there should be less variance.  It is a big pool problem for part time miners.

The solution to this is simply have CGMiner do load balancing and mine on multiple pools.  The only problem would be that your income would be spread out but for people who are into mining this should not be an issue.  You could mine on the #2-#6 pools and approach a virtual 40% pool.

[Billbags]

This link is a good comparison of BTC Guild and other pools over the last several months. (Note that they are comparing p2pool NOT p2pool.org)


https://bitcointalk.org/index.php?topic=416933.0

[stryker]

Cheers for the link, I'd not seen that thread

[DrG]

This link is a good comparison of BTC Guild and other pools over the last several months. (Note that they are comparing p2pool NOT p2pool.org)


https://bitcointalk.org/index.php?topic=416933.0

That's the point of what I was saying about load balancing.  If you spread the hashrate over multiple pools you can decrease your variance - in that example you would be mining at P2Pool, Eligius and BTC Guild all at once.

[Syke]

https://www.btcguild.com/index.php?page=pplns

Approximate Pool Luck* (24H / 3D / 1W / 2W / 1M / 3M / All Time): 85.603% / 95.163% / 93.237% / 91.922% / 88.917% / 92.513% / 98.034%

It's hard to find a recent time period where BTCGuild didn't have bad luck.

[xstr8guy]

https://www.btcguild.com/index.php?page=pplns

Approximate Pool Luck* (24H / 3D / 1W / 2W / 1M / 3M / All Time): 85.603% / 95.163% / 93.237% / 91.922% / 88.917% / 92.513% / 98.034%

It's hard to find a recent time period where BTCGuild didn't have bad luck.

Look again today. Luck is pretty great right now.

Approximate Pool Luck* (24H / 3D / 1W / 2W / 1M / 3M / All Time):
131.043% / 104.127% / 93.414% / 93.695% / 91.498% / 93.141% / 98.106%

[notlist3d]

https://www.btcguild.com/index.php?page=pplns

Approximate Pool Luck* (24H / 3D / 1W / 2W / 1M / 3M / All Time): 85.603% / 95.163% / 93.237% / 91.922% / 88.917% / 92.513% / 98.034%

It's hard to find a recent time period where BTCGuild didn't have bad luck.

Look again today. Luck is pretty great right now.

Approximate Pool Luck* (24H / 3D / 1W / 2W / 1M / 3M / All Time):
131.043% / 104.127% / 93.414% / 93.695% / 91.498% / 93.141% / 98.106%

I was a long time user of this pool, i still like it and have as a backup.  Sadly look at the 1 week plus.  3 good days does not make up for 3 month's of bad luck - 93.414% / 93.695% / 91.498% / 93.141%.  Granted it's not horrible luck still in 90's but i still like the overall 98 percent much better then past 3 months.

[xstr8guy]

https://www.btcguild.com/index.php?page=pplns

Approximate Pool Luck* (24H / 3D / 1W / 2W / 1M / 3M / All Time): 85.603% / 95.163% / 93.237% / 91.922% / 88.917% / 92.513% / 98.034%

It's hard to find a recent time period where BTCGuild didn't have bad luck.

Look again today. Luck is pretty great right now.

Approximate Pool Luck* (24H / 3D / 1W / 2W / 1M / 3M / All Time):
131.043% / 104.127% / 93.414% / 93.695% / 91.498% / 93.141% / 98.106%

I was a long time user of this pool, i still like it and have as a backup.  Sadly look at the 1 week plus.  3 good days does not make up for 3 month's of bad luck - 93.414% / 93.695% / 91.498% / 93.141%.  Granted it's not horrible luck still in 90's but i still like the overall 98 percent much better then past 3 months.

As of this moment, week+ is looking a bit better.

Approximate Pool Luck* (24H / 3D / 1W / 2W / 1M / 3M / All Time):
128.912% / 117.454% / 103.558% / 96.538% / 93.610% / 93.756% / 98.189%

[philipma1957]

variance!! do I win a prize? 

Ah... regarding this hearsay about a large hasher withholding blocks, I was not aware of that but as I said at the beginning of this thread I was starting to wonder if it was something like that going on.

Don't get me wrong, I love the pool's site, best interface around and very clear, plus the operators at least appear to be very transparent.  It was just that terrible luck they had a few days ago that left me thinking WTF??

Well the pool was running close to 105.9% luck if I recall for their 3 month average before the sh*t hit the fan in May (when the withholding attack started supposedly).  At that time nobody complained that the pool might be cheating, although people did allege that in the past back in 2011.

So when it's up almost 7% (106-99) nobody complains.  When it's down 7% something is wrong.

The withholding attack is not hearsay BTW, the person was identified.  But he states his custom version of CGMiner was at fault.  The thing is it could have happened at any pool, and some pools don't even bother with the nice stats. Too bad the withholding didn't go on at ghash.io lol.  The same "attack" also occurred on Eligius and the guy is asking for payment which Eligius held once they discovered the problem.  You can check the pools subforum for this info.

Eleuthria might not be eligible for sainthood but he's one of the best pool ops around.

The problem with the withholding event is very simple it now is true it can be done.

If anyone with a copy of that custom cgminer wants to attack any pool and make it hard to detect they can.  Since cex.io is so big they have the ability to toss 2ph to 4ph in the shitter on any pool they want. 

the only defense for a with holding attack is solo mining.