Re: [400GH/s] p2pool: Decentralized, DoS-resistant, Hop-Proof pool

I hope you all find this interesting: NPW 5.1 p2Pool - bad luck or flawed?
Let me know if I've made any egregious errors in how I've described how p2Pool works.

Nicely done, but there is NO explanation WHY it is 10% longer than it "should" Tongue

And I have absolutely no idea if it is even possible or how.

I have a idea, but unsure how to prove it.
In my case P2pool.info ALWAYS and sometimes even my local node shows more hash rate than I have on my miner.
Maybe it it all ok in pool itself, just numbers send/taken to analysis are screwd? By this magic 10% ?

hashrate by p2pool on your PC and on p2pool.info is a GUESS based on the shares submitted at that time. I've had it say I've got 40gh/s because I mined with 1500 diff shares and got 5-10 shares in quick succession.

Maybe this "guessing" is the troublemaker. Some missed comma in code or something and all calculations can be trashed.
IE it is calculating that diff 600 share need 270MH (numbers form the roof) but in fact it need only 250. This way pool "guessing" that pool HR is higher than it really is, so time to solve block is shorter. But in fact it takes longer because real HR is lower.
In "normal" pools where SD=1 error margin is much lower (higher accuracy), we have SD~600 and rounding numbers up/down can make statistic troubles.

Edit:
From my week graph:
Local rate = 132MH/s
Local rate reflected in shares = 164MH/s
So pool "guessing" that I have 20% more hash power than I really have.
If everyone have +10% on stats then "bad luck" problem is only a statistic myth. If more ppl have over +10% ... then everyone should mine there Tongue