Quote from: kano on March 04, 2016, 10:09:45 PM

Extra work restarts caused by alt-coin merged mining block changes invalidating valid BTC work.

At Bitminter only a new bitcoin block makes old work get rejected.

But yeah, have a look at the worker options and disable "Merged mining stale flush". That should reduce the frequency of those.

There are still some special cases where the server will ask the client to immediately switch to new work. For instance if the server gets a transaction with an extremely high fee.

Quote from: Icon on March 04, 2016, 09:52:58 PM

Well one thing you forgot to mention is using a high(er) diff rate, means when you finally submit your work and the time it takes to get there, the data could be old and get rejected. Thing is with a higher diff rating all that work for all 64 shares will be rejected..

And when accepted, all 64 are accepted. Higher diff just means higher variance, both for accepted and rejected work. Remember that you are equally likely to be lucky as you are to be unlucky. Meaning the extra variance could cause a higher accepted count than usual, and lower than usual rejected.

Over time both will even out though.

Quote from: Icon on March 04, 2016, 09:52:58 PM

Did you change from Proportional Payment mining to full PPLNS? No more not getting paid for older shares?

I don't understand what you mean. The pool started as proportional in late june 2011. After a couple months we switched to PPLNS with shifts.

Bitminter was the first pool to use PPLNS with shifts. A couple other pools had done PPLNS (without shifts) before that. I don't think any of them are still around now. That was in the early days when many different reward systems were being tried out.