It occurred to me this morning that the Exit Poll Discrepancy could be fully explained in terms of voter disenfranchisement -- that is, long lines, intimidation, and discarded provisional ballots in Democratic-leaning precincts.
Here is a simplified example to make my point:
Assumptions:
(1) The state of Flohio has 10,000 registered voters, of whom 5000 live in the city and 5000 live in rural areas.
(2) Of the city voters, 70% support Kerry and 30% support Bush.
(3) Of the rural voters, 32% support Kerry and 68% support Bush.
(4) Exit polls are set up at one city precinct and one rural precinct, and the results of those two exit polls are averaged to produce the statewide exit poll data. The two precincts are weighted equally, because the exit polling company determined that voter turnout was consistent across rural and city areas in 2000 (say, 66% turnout in both areas) and they predict the same for 2004. So, the exit poll predicts 51% Kerry and 49% Bush.
(5) The voting machines in Flohio are flawless and the votes are counted correctly.
(6) In the year 2004, long lines and voter turnout lead to reduced voter turnout in the city, but overall the voter turnout is increased. Specifically, voter turnout is 74% in rural areas and 64% in the city.
The actual tabulated votes are:
City votes for Kerry: 5000 * 70% * 64% = 2240 votes
Country votes for Kerry: 5000 * 32% * 74% = 1184 votes
City votes for Bush: 5000 * 30% * 64% = 960 votes
Country votes for Bush: 5000 * 68% * 74% = 2516 votes
Bush gets 3476 out of 6900 votes, or 50.4%
Kerry gets 3424 out of 6900 votes, or 49.6%
Conclusion:
It would be a worthwhile pursuit to attempt to quantify to what extent voter disenfranchisement altered the results of the election. One way to do that would be use the precinct-by-precinct results from 2004, and "normalize" them by doing calculations based on the year 2000 voter turnout. If the results are in line with the year 2004 exit polls, then perhaps the recount efforts are barking up the wrong tree.
I believe both factors (improper/fraudulent counts and voter disenfranchisement) played a part in skewing the results of the 2004 election, but I'm not sure which was the bigger factor. To my way of thinking, either one should be sufficient to nullify the election.
For more information about voter disenfranchisement in 2004, see these two excellent articles:
"Kerry Won Ohio: Just Count the Ballots at the Back of the Bus"
http://gregpalast.com"The Perfect Election Day Crime"
http://www.tompaine.com/articles/the_perfect_election_day_crime.php