A year ago I analyzed a mining pattern that suggested a correlation between Satoshi computer and another computer mining at the same time. This relation had been highlighted to me by a another user in the forums (I don’t remember who) . From this correlation I conjectured that the same computer was mining both patterns or two computers were under the same heavy CPU load at the same time. I was wrong. This is the graph:
A week ago I saw the graph again and realized what is the real cause of such apparent correlation. The cause is that the extraNonce is incremented not only when time passes but also when a new block is solved, either by the miner or by any other miner. This is because the outer mining loop increments the extraNonce at the beginning. When the best chain is updated, the inner loop (which increments the nonce) is stopped and the outer loop is executed.
So when Satoshi stops mining, on July 30th, every other miner seems to slow down. Also on July 4th, when Satoshi goes back mining, the remaining miners seem to increase their hashing power. But now I know this is a false impression.