Assuming that the error is evenly distributed and that there is no fraud in the hand count. You can pull the exact machine counts for each of the three counties ( cnn.com ) and note that as of this posting, Broward has counted 277 of 609 precincts, with Gore picking-up 68 votes. One must assume that on average, there are the same number of voters in each counted precinct as each uncounted one and that on average the counted and uncounted approximate the final Gore/Bush vote ratio. When you do the math for the other two counties, using the second machine count for a basis, Bush should have a 719 vote margin when the hand count is over. This assumes no fraud and that the machine to hand count tendency remains constant. The exact machine vote count for each county must be used, because both parties pick-up votes due to incompletely punched ballots. We assume that both pick-up votes at the same rate, but Gore has many more votes in these counties, so he gets the net gain. It is possible that my math was faulty, but the basis is sound. A major deviation from this is fraud, or the assumptions are wrong.