My reading of the 1.2 value was it being without qualification as in: WCG expects to send out 1.2 results average to volunteers to net 1 result that can be transferred to the Master Science Database via the Assimilator.

I'm not sure if this was a first for any BOINC life science project to put a mini WU test into the jobs. I expect for extreme prudence to be maintained, since the computing continues to be outwith the immediate control of the scientists, which is why quorum 2, 3 even 4 was run in the early days of BOINC to get absolute verification.

This is also why HCC is not suited for zero redundancy running. Each of the 90 million images needs 100.00000% perfect confirmation for the reference database to have any meaning as a source for further studies. No wet lab test to confirm the usefulness of a candidate compound that was identified out of the millions of tests.

We have dug through the code to figure out what was going on with these and we found that we had a slight bug that was allowing results marked invalid to get a second chance at validation when additional results were returned. This is why results marked 'invalid' were turning into results that are 'valid'.

We have several checks in place that look at the results that are returned on their own. Some of these checks are used to immediately determine that the result is invalid. Others are used to determine that we are not sure about the result and a second copy needs to be sent (and thus the result is marked inconclusive). In the two cases above, the results failed one of our checks that marked the result immediately invalid. However, as it has been clearly shown (and we have manually confirmed) that this check is sometimes incorrect in determine a result to be invalid, we are changing that check to mark the result as inconclusive in the future. This will result in another copy being sent to confirm the answer.

Kevin, thanks for the detailed feedback !!

Great!

Thank you for the feedback. Jean.