Weekly pool and network statistics

[eleuthria]

Nice graphs. Look at BitMinter go! How do you like them apples?

Why the wonky graph for Slush? I thought his pool wasn't hoppable anymore.

Nope, slush is still very hoppable at the moment.

[organofcorti]

Nice graphs. Look at BitMinter go! How do you like them apples?

Why the wonky graph for Slush? I thought his pool wasn't hoppable anymore.

As eleuthria writes Slush is still hoppable - the promised change to DGM hasn't occurred yet.

Proportional pools are hoppable by submitting shares only until the pool has received a certain amount of valid shares in total, so the second group of charts shows the current proportional pools' average hashrates per round vs total submitted shares / Difficulty for that round.

Two of the standard proportional pools and both of the exponentially scored proportional pools show significant decreases in hashrate for longer rounds, indicating a higher amount of pool hopping hashrate added early on in a round.

[organofcorti]

I don't know that I would classify it as directly related to DGM, it just counts the block as zero shares, thus bypassing DGM all together and paying out the current score (which is basically the same as the previous block score).

I think you're right - as long as the expected amount of each share under your reward method has an expectation of p*B and all shares and rewards are still accounted for, then it shouldn't matter.

[Inaba]

Yep, that's exactly what happens.  Basically runs through the block payout algorithm twice in a row while attributing all the shares to the previous block.

[organofcorti]

Apologies to BitcoinPool and their miners - the pool's website was down today and I was unable to retrieve the weekly data.

I have had some feedback indicating that 50BTC.com should not be included in the pool hashrate stats due to the unavoidable error caused by the minimal published data the pool makes available. At the current hashrate of 2000 Ghps, the standard error would be +/- 8.8% and will reduce id 50BTC's hashrate increases as a proportion of the network hashrate. Anyone else have any thoughts on the matter?

I've added some more charts, and I've added a guide section for the charts.
Chart 1: List of all pools with public data and their various statistics averaged for the last seven days - for smaller pools the average may be more or less than seven days, depending on number of blocks solved for the week.
Chart 2: Network hashrate, hashrate of the largest mining pool, combined hashrates of the three largest mining pools, and a line representing 50% of the network hashrate. Handy if you're worried about 51% attacks.
Chart 3: Chronology of pool hashrates, averaged per week.
Chart 4: Average hashrates per pool per round for the week.
Chart 5: Chronology of pool luck indexes, averaged per week.

Chart 6: Chronology of round length divided by difficulty, averaged per week.
Chart 7: Percentage of orphaned blocks produced by the pool, averaged per week.
Chart 8: Hashrate vs round length for hoppable pools (the larger the hashrate increase at the start of a round, the larger the loss to strategic miners).


Highlights:

• It seems that the thought of actually supporting their pool financially had led some Ozcoin miners to leave, losing the pool 300Ghps and 2% of the network hashrate.
  
• p2Pool still hasn't seen any bad luck for six weeks in a row.
  
• Deepbit finally gained a little in terms of network hashrate, and has been extremely lucky for the last seven days.
  
• Hoppable pools apart from Deepbit are being hopped for a significant portion of the initial hashrate.

Weekly average pool statistics, 19th August:





















[/list]

[wknight]

As always.. great job!

[DrHaribo]

Wow, awesome charts!

I had no idea there was still so much pool hopping. You can clearly see the pool hopping effect, even on Deepbit. Pool hopping on Slush's pool seems massive. I had no idea.

Why do people still mine there and lose so much income?

I guess Deepbit miners think "I only lose a couple percent to pool hopping, it's no big deal." But the loss from mining 24/7 on the other hoppable pools look like it is nothing short of massive. Pool hoppers put a serious hurting on your wallet.

[Subo1977]

P2Pool rock's 

[organofcorti]

26th August weekly pool statistics

Once again Bitcoinpool's website was down, although it had been up for a few days during the week. This can't be a good sign for Bitcoinpool or Bitcoinpool's miners.

I have added 50BTC.com. The average hashrate attributed to this pool cannot be exact (as discussed previously). I have decided that the small error that could result from the estimation of hashrate from blocks produced will, in the long run, be negligible.


Highlights
* Pool added: Bitparking merged mining pool.
* All charts now published at organofcorti.blogspot.com to allow easier searching through the thread and also allows me to post higher quality easier to read charts.
* EclipseMC was the only pool to see a significant increase in edit: network hashrate this week.
* p2Pool is now up to 8 weeks in a row during which the averaged luck for the week has been average or better than average.
* Including 50BTC.com brings the top three pools' combined hashrates much closer to 50% of the network hashrate.
* No slowdown of pool hopping. I hope by next week to add a table with an estimation of how much fulltime miners lose to pool hoppers (details on how to calculate this are in NPW 6.2: Bitlc)

[wknight]

Great job as always!

[racerguy]

p2pool = the little pool that could.

[organofcorti]

2nd September weekly pool statistics

Bitcoinpool is back up this weekend, and Coinotron just increased their hashrate by an order of magnitude so they're back in as well.

Highlights:
* Check out the effect Bitminter's race had on their hashrate per round stats - that's a big spike there, thank you very much fefox and GPUMax.
* Many of the big pools are lucky this week.
* Full charts at organofcorti.blogspot.com

[Graet]

as always - excellent
but
which are the bug pools
or have you adopted a kiwi accent lol
G

P.S. I know - it is a rare typo

[organofcorti]

as always - excellent
but
which are the bug pools
or have you adopted a kiwi accent lol
G

P.S. I know - it is a rare typo

Argh! I couldn't think of a good come back! I've been off work sick the last few days - that can be my excuse.

Thanks for pointing it out. Although it was a kind of funny typo.

[Graet]

as always - excellent
but
which are the bug pools
or have you adopted a kiwi accent lol
G

P.S. I know - it is a rare typo

Argh! I couldn't think of a good come back! I've been off work sick the last few days - that can be my excuse.

Thanks for pointing it out. Although it was a kind of funny typo.

I did lol, and illness is a good excuse - used it a few times myself over the last few days :/
hope you feel better soon

[organofcorti]

8th September weekly pool statistics

Mt Red's website was down tonight, so no data for them. Sorry Mt Red miners. All other pool websites were up.

In other news, I've added a new chart for measurement of pool luck - the negative binomial lower tail CDF probability of the number of shares submitted resulting in a number of rounds (actually, the Gamma distribution approximation of this). A probability close to 0 mean the pool was very lucky;  a probability close to 1 means the pool was very unlucky. The advantage of this method is that unlike the other two methods, variance is not being measured so pool hashrate does not have an effect on the result. The luck of a large hashrate pool will be directly comparable to that of a smaller hashrate pool.

Since I think this is a more useful measurement of luck when comparing pools, it is at position 5, above the other two luck measurements. You can find this new and exciting chart at http://organofcorti.blogspot.com.

[doublec]

I see Bitparking took the hit for everyone's good luck over the week!

[organofcorti]

I see Bitparking took the hit for everyone's good luck over the week!

Yes, that was such a strangely lucky week for so many pools and strangely unlucky week for Bitparking that I had to check to make sure my scripts weren't doing something weird with the difficulty calculations (well, not for Bitparking, since doublec provides the difficulty at which a block was solved in the block history, which is nice). But blockexplorer's nethash looked up to date, so it just seems to be a week of asymmetric luck.

The new negative binomial CDF charts at organofcorti.blogspot.com are a better measure of luck imo since there are no variance changes due to difference in pool hashrate. On these new charts pool luck is much easier to compare between pools of different sizes.

[organofcorti]

Updated with correction to Bitminter - should only have been 1.4% orphans.

[organofcorti]

16th September weekly pool statistics

Good news, everyone! All websites were up this weekend.

Itzod continues to grap a greater proportion of the network hashrate and seems to be doing quite we.. Ozcoin's hashrate has clearly levelled out - it's interesting how very stable their hashrate seems to be now that proxy pools are using them less.

Another interesting change is DeepBit's average hashrate per round as a function normalised shares per round (Chart 9). They are being pool hopped a litte less, but much more interesting is the fact that they don't seem to have a decrease in hashrate at the start of a round anymore. I wonder what changes they've made to the pool servers recently?

You can all charts at http://organofcorti.blogspot.com.