Did I ask you to change anything? No. Simply stating the facts.
I didn't say you did. I just asked you to allow us to run our pool the way we choose to.
While not holding data to prove one way or the other, I must concur with jgarzik for the simple fact that unless there is something predictable about sha256, and thus we should all be working towards replacing it for bitcoin usage, the random nature of the calculation means that historical data bears no weight into future prediction.
This is just another way to say that just because you've gotten heads 10 times in a row, doesn't mean your changes of getting heads for the 11th time is bigger, assuming a well balanced coins and a good, strong throw. It is always 50%, really.
If you can in fact prove that you found a pattern on the hash calculation towards H==0, that is a potential attack vector and should be investigated. Please share the real world data you speak of!
http://bitcoinpool.com/thedata.html3 different datasets and a comparison of the number of shares found for 3 different askrates to those that would have been found at lower askrates or with CPU miners.
It isn't proving the discovery of any pattern aside from the fact that the likelihood of finding an answer in 1 second on a CPU miner is < 5%, whereas it causes a significantly higher server load to try.
Even if a new block is solved every 600 seconds, setting an askrate of 20s or more would significantly decrease server load, and not likely increase the chance of an answer being for the previous block by any significant amount.
I'm not saying any pattern has been discovered. I never said SHA256 has been broken. I am however showing you actual proof that it is not a 50/50 chance on whether you will find an answer immediately upon switching to a new getwork.