In yesterday's post I gave a number of specific examples of how messed up mail ballot accounting was in the bungled election of 2004. Now we'll see how thoroughly messed up it still was in the bungled elections of 2005 even after Deanron promised sweeping fixes.

First recall the bungled 2005 primary, which, as in Nov. 2004, was certified on the basis of a bogus mail ballot report with fictional numbers that didn't even add up. (the Mail Ballot Report itself is here). The canvassing board apparently didn't notice that the report showed that 390 ballots were uncounted and unaccounted for and Deanron's staff apparently did not volunteer any explanation. Instead, they handed out a 40 page print-out of a "Batch Accountability Spreadsheet" purporting to prove that all ballots were properly accounted for. The actual spreadsheet is posted here. The first clue that it is a worthless accounting tool is that none of the columns are summed. Had the columns been summed, the diligent ballot accountant would have noticed that the numbers didn't add up. Had the canvassing board and the media been given the actual spreadsheet on certification day, the public would have realized that the numbers didn't add up. But a 40-page print-out without summed columns is a terrific tool for a snowjob. I later asked for a copy of the spreadsheet. Even though Deanron's people could have emailed me the spreadsheet within minutes, it took them a month to make me a CD for which they charged $12.50.

Now that I have the spreadsheet and what purports to be a complete transaction log of absentee ballot processing, I can truly appreciate how bogus the Mail Ballot Report really was and how cheesy the whole "ballot accounting" system really is. Gory details follow --

For a review of basic technical concepts and vocabulary, see yesterday's post.

(1) The Mail Ballot Report probably understates the number of actual returned ballots by more than 1,000. The Mail Ballot Report asserts that 251,480 ballots were returned, but also acknowledges that said number was not really the number of ballots returned, but only the sum of ballots counted + ballots rejected. According to the transaction log, the number of returned ballots was 252,313. But that does not include the "Wanda rejects", of which the spreadsheet asserts there were 196, nor the Address Confidentiality Program (ACP) ballots, which the Mail Ballot Report shows 24 (that imputes a total of 252,513 returned ballots). For some reason, the Mail Ballot lumps the Wanda rejects in with other misc. rejects and reports a total of 57.

(2) The Mail Ballot Report asserts that 3,526 ballots were "returned too late" and "Reported by DIMS and reconciled with a handcount". According to the transaction log, that number should have been 3,817. It turns out the remaining 291 had arrived but were not entered into DIMS until shortly after the Sep. 30 certification meeting [example]

(3) The total in the "Cards Cast" (ballots counted) column in the spreadsheet is 245,651 or 10 more than the 245,641 that was reported in the Mail Ballot Report

(4) It's hard to interpret what the other columns in the spreadsheet actually mean, but here's what we can tell: The total ballots verified and eligible to be counted is 246,144. That is close, but not identical, to the Mail Ballot Report's claim that there were 246,114 DIMS voters + 24 ACP voters. The Mail Ballot Report also claims that 21 envelopes were empty and that the canvassing board rejected 85 ballots as spoiled. This leaves 246,038 ballots that should have been tabulated. But if the spreadsheet is correct, then only 245,651 were tabulated, meaning that 387 ballots disappeared. According to news reports, 18 of these ballots were discovered uncounted a week after certification. Nothing in the "accountability" spreadsheet that I can see would have identified this group of uncounted ballots.

My hypothesis from looking at some of the other numbers in the spreadsheet is that a stack of the rebatched ballots were overlooked and didn't get counted. That's consistent with an insider tip I received on the eve of the November 2005 election. Deanron flatly denied that any ballots were left uncounted and insisted that the discrepancy was attributable only to "data entry" errors. But looking at all of the data, that flat denial seems completely bogus. If it is not completely bogus and data entry errors could cause 400 phantom envelopes to appear at the front end, then the entire ballot accounting system is completely bogus. And this bogus accounting was for a dinky off-year primary where the number of returned ballots is only a quarter of what we can expect if the next Presidential election is entirely by mail.

Coming up: more bogus ballot accounting in the November 2005 election.