Greys Harbor was really strange. They double counted votes and my figures show they lost 2240 votes when they corrected the double counted votes. After the corrections, Greys Harbor swang from Rossi's favor to Gregoire's.
It did switch the winner in the county, but the Rossi percentage only fell from 50.4% to 49.6%, which is not an implausible difference for double counting an area that was slightly pro-Rossi ( about 3989-3594). Just as with the "mysterious" Ohio precinct that had an extra 4,000 Bush votes, it really does seem to be a clerical error that was caught. The original GH count would have given them the highest % turnout in the state by a wide margin, so it should have been obvious that there was some error at work.