Bitcoin Forum

Noticed Bitminter went down, then noticed this while investigating. (https://i.imgur.com/c6USUWI.png)

7:09PM MST, May 18th.

This is much more accurate:

http://blockorigin.pfoe.be/chart.php

And BTC Guild have deliberately throttled their share to just under 40%.

Ahh, OK.  I'd seen them with a high percentage before but never that high.   Disregard.

Unless BitMinter went down over 2k blocks ago, I'm not sure how you could say it's more accurate. If it's not accounting for the outage quickly, it's obviously going to be pretty inaccurate. Idunno the situation, though, and know El has measures in place to stop BTCGuild from achieving too high a percentage of hashpower...

Ho-hum.

Remember a year or two ago when Deepbit ruled the pools, and touched - even occasionally surpassed - 50%?

It's a culturally self-correcting 'problem'.

Also, there is zero incentive for a pool to attempt a 51% attack - an attack which can't ever be anywhere near as profitable as running the biggest pool.  Not even close.  Maybe BTCGuild could be sandbagging another 10% of the network hashrate with a bunch of ASICs it's got hidden by running them on other pools.  So?  They bring them on line, add them into their 50%, attack the network, and maybe - maybe - they get six or eight blocks.  Big whoop.  That's worth years of past work; and all of their future profits?

Silliness.

oh look, it's another "omfg we're close to 50%!1!1!" thread.

see:

Need more p2pool...

Yea, just because a 51% pool could happen doesn't mean they would use that power for evil for the reasons listed above.  But as others had brought up, if the pool fell into the wrong hands with a hack, then the hackers could cause some serious issues for the lulz.

In more words:
Kinlo's block origin is more accurate than blockchain.info about the pool percentages for block hashing, because block origin uses their published results whereas blockchain.info collates based upon who propagated each block to them.

Thanks.  :)

Just an aside:  Blockchain.info is almost always accurate on BTC Guild's blocks since we always use the same payout address and have a coinbase message.  I'm not sure which they're looking for (or both), but as a result I can't think of a block reported as BTC Guild that came from another source.  Their 24 hour chart is not "inaccurate", BTC Guild really has been floating in the 47-49% of blocks for the last 24 hours most of the day due to extremely good luck (look at our PPLNS stats, the last 15 closed shifts have ended up paying 10-60% above PPS).

The reason blockorigin.pfoe.be is always linked and people told not to rely on blockchain.info is that a 24 hour chart is so heavily influenced by luck that it means very little.  If pool A has good luck, they will show up big.  If pool A has good luck while B and C have bad luck, pool A looks HUGE.  If blockchain.info would fix their 48-hour/4-day charts (they only marginally increase the span, nowhere near 2 days/4 days), and perhaps defaulted to the 4-day chart, it wouldn't be quite so annoying.

BTC Guild has only barely gone above 40% during any 2016 block period, never hitting 41% as best I can tell [it may have, but if so it was very brief even on a 2-week average].  We've slowly been moving downward after the recent changes to PPS fees and forcing getwork users to PPLNS, which was the reason those changes were made in the first place.

Thanks for the deeper insight!

For the past week:

(keep in mind it's the proportion of blocks solved, not the actual hashrate, that is important).

http://4.bp.blogspot.com/-ZvlzshcFRV8/UZh7oIHuRJI/AAAAAAAAICc/yeqIGkSUI6k/s1600/1.pie.19-05-2013.png (http://organofcorti.blogspot.com/2013/05/19th-may-2013-weekly-pool-and-network.html)

Whether fortunate or unfortunate BTCGuild has been having bad luck for the last 4 weeks (sans minor outbreaks like earlier on 5-18).  If they had good luck then their solved blocked would be considerably higher and then it might matter.

Ultimately it comes down to the miners to have the common sense to not over-concentrate things.