I hear what you are saying.... I love the fact that Qt is using real statistics to calculate the Fees.
It is probably still worth the time and effort to always look at:
http://bitcoinexchangerate.org/feesbefore sending a transaction...
Also.. be away that the way Qt computes fees may depend your it's copy of the mempool. This means if Qt hasn't been running all day, or if it isn't fully synced, that the stats may be off.
Your are not answering my question: did Bitcoin Core 0.11.0 changed something in how Smart Fee is calculated?
Thanks for your answer.
Thank You for the rest: free info is always valuable.
I'm proposing an answer, but refusing to commit to it based on the sparsity of the information provided.
[Q1] - What block height were you at on 0.10.2 when you started the send transaction dialog?
[Q2] - What was the size of your mempool on 0.10.2 when you started the send transactions dialog?
[Q3] - How many incoming / outgoing connections did you have on 0.10.2 when you started the send transactions dialog?
[Q4] - How long had 0.10.2 been running with active connections when you started the send transactions dialog?
[Q5] - What block height were you at on 0.11.0 when you started the send transaction dialog?
[Q6] - What was the size of your mempool on 0.11.0 when you started the send transactions dialog?
[Q7] - How many incoming / outgoing connections did you have on 0.11.0 when you started the send transactions dialog?
[Q8] - How long had 0.11.0 been running with active connections when you started the send transactions dialog?
The answer could be either that 0.11.0 botched the fee calculations, ...OR... your 0.11.0 client instantiation was making a calculation based off of less network information, or less congestion than your 0.10.2 client instantiation.