why we get so often Orphaned blocks at this times?
It has to do with the pool code, if it's optimized to get solved blocks and work out quickly, block change time, internet connectivity/latency of their mining operation.
Kano and Ck have done a lot of work on their pool code to avoid orphans and empty blocks, you could always try mining on a pool run by people who actually care about the bitcoin network:
Blocks are lost a few ways:
1) The obvious network orphans - fewer means a better connected pool
2) The hidden pool orphans when the bitcoind test says it got a network block, but the pool got it's block just after that, but before the pool software knew about the network block change (these show up on our pool - our 2nd orphan was like that - but most pools don't show these)
3) The lost stale blocks due to processing of work by the pool being too slow after a network block came in - due to slow pool processing
4) The lost stale blocks due to the miner getting the block to the pool after the pool knows about the next network block - due to a slow miner network, a slow pool network or slow pool processing
Just to let miners know, Kano and I worked on minimising the latency of bitcoind under the current flood of transactions and I'm happy to say that we've been able to deploy some of my custom modifications to this and solo ckpool which still guarantee maximum transaction processing (i.e. never a zero transaction block) while minimising the risk of stale work / orphans. Plus we are NOT using a transaction blacklist; all transactions are prioritised on their own merit as per bitcoind rules.
