I just fielded a question or two about the apparent 'double' scores on 30-07-2010.

To help answer the same question for anyone else, the dates on the "Last 28 Days" table are calculated labels on the table in the web page (eg today, today -1, -2 etc).

Because the non-BOINC process froze and did not fail, the nightly rollover was prevented from shifting the data across in the daily data table, which in theory should have created a column full of zeros because the refresh did not happen on 29th.

Consequently the table appears to be out by a day prior to 30-7-2010.
The date label saying 29-7-2010 contains the volumes from 28-7-2010.
The date label saying 30-7-2010 contains the volumes from 29-7-2010 and 30-7-2010.

The volumes are correct.
No data is missing.
No data has been duplicated.
The 'apparent' anomaly is self correcting and will shift itself out of the "Last 28 Days" table day by day.
This does not affect the weekly or monthly stats.

Trust us. It is Auto-magic