Post
Topic
Board Pools
Re: [CLOSING JUNE 30] BTC Guild - Pays TxFees+NMC, Stratum, VarDiff, Private Servers
by
man114
on 16/06/2015, 06:12:27 UTC
Another victory for centralized mining.
Finally, I have been growing concerned for some time now about attempts to defraud pools. The pool's luck has been on a decline for over a year. The luck on a few other pools has also shown a negative trend. While it is not impossible that it's a coincidence, this is something I have been constantly made aware of and am helpless against. There is no way to know whether it's just bad luck, a small bug in older miners (BTC Guild probably has the highest percentage of first/second generation ASICs) resulting in a few % of block-solving shares to disappear, or a large pool trying to hurt the competition (many of the largest pools have large private mining operations now). It would only take a fraction (1 PH/s or less could do it) to cause significant harm to a competing pool, and that activity could be masked by proxies and multiple accounts to be impossible to catch.
What do you mean with this?

I mean the pool's luck for the last 3 months has been 88~89% of expected block solves.  Eligius' 3 month luck is almost identical to that last I checked in their thread.  Slush doesn't give a 3 month luck but his 1-month luck is pretty close to that.  I do not know if that is statistically bad enough (Guild + Eligius pools in aggregate) to point towards something other than variance.  Both pools combined are only about 5% of the network, so 12% under expectation over 90 days might be something that could happen 1 in a hundred samples.  I'm sending organofcorti a message about it since that's his area of expertise.

88-89% doesn't seem anything out of the ordinary for variance for a 3 month period. Slush has had a bad run of luck in the last two weeks or so and the drop in hash rate certainly hasn't helped. Of course the drop due to summer certainly won't help hash rates any.

I don't know if older ASICs would have anything to do with it, an old 1th/s is going to do the same but use more electricity to do it.