Thanks for the link, Dave.

The answer from knreed couldn't be more cryptic but I guess what he wants to say (should have said) is that one server (the result server that responds the current total back to the client) keeps an exact (float) total while the statistics (web) server that serves "My Grid" actually adds integers to the online total each time a result is sent. That's quite weird and I can't possible think why some scientist would ever do this (summing rounded figures instead of rounding a sum) as it's totally wrong. It means that in worst case the grand total of received points by any user can be off by half as many points as the total number of *results* he sent (as he can win or lose 1/2 point for each result he sends). His client will show the correct value but we all look at the statistics and I guess the MOT totals are subject to the same statistical error. Of course statistics also say that the effect levels over time as it's a symmetrical error but *theoretically* our MOT total *could* be off by 80000 points (we all sent some 160000 results) in either direction. 1/2 point in some 200 (per result) is only about 1/4% off and (again!) this is the WORST CASE statistical error but it still is a (huge) methodical bug.


Best regards, Stefan.