Can we forget about John Ashcroft and go back to the Florida recount for a moment? The press has long since left the theater, and the final credits have almost finished rolling, but there's another plot twist to the story. A pretty big one.
I'm referring to the "press recount" conducted last week by the Orlando Sentinel in Lake County, a fairly small, 90,000-vote county in central Florida that George Bush carried by 15 percentage points.
You wouldn't expect a Lake County recount to reveal many new votes for either candidate. After all, the county uses the supposedly more accurate optical-scanning voting system, in which voters mark their ballots with a pencil--no chad-producing "punch cards." What's more, the 3,114 ballots examined by the Sentinel were "overvotes"--ballots the optical scanning machines had rejected because they detected marks for more than one presidential candidate. If you followed the coverage of the recount, you know that overvotes were not a central focus of the Florida fight. Instead, the recount controversy centered on "undervotes"--ballots on which machines detected no presidential vote at all. It was undervotes that Gore was desperately trying to get manually counted and that the Florida Supreme Court had ordered counted before that count was stopped by the U.S. Supreme Court.
What could overvotes yield, anyway? If a ballot is marked for two candidates, it's irretrievably spoiled, right? True, some people--including Bush lawyers seeking to discredit an "undervote-only" recount--raised the possibility that some overvotes might be salvageable if, say, voters actually wrote "I want Bush" on their ballots. But this possibility seemed almost theoretical. "There's nothing in the record that suggests there are such votes," Gore attorney David Boies asserted confidently when asked about the possibility in oral argument before the U.S. Supreme Court.
We now know how wildly off-base Boies was.
We know because in Lake County the Sentinel examined 3,114 overvotes. And one-fifth of them contained exactly the "write-in mistake" that Boies had dismissed as nonexistent. More perversely, the majority (376) of these ballots were clear votes for ... Boies' client, Gore. "In each case, an oval next to his name was filled in with a pencil and the voter mistakenly filled in another oval next to a spot reserved for write-in candidates, writing in Gore's name or running mate Joe Lieberman's there as well," the Sentinel reports. Some 246 ballots contained the same Write-In Mistake, except that the voter both marked and wrote in "Bush." But, all told, Gore would have gained 130 votes in this one measly little county had its overvotes been manually tallied.
In retrospect, it seems glaringly obvious why voters would make the Write-In Mistake. If you're a first-time voter, after all, and you see a ballot that says "Mark your candidate" and then another line that says "Write-in," you might easily think that the latter phrase was instructing you to write in your candidate's name--just to be sure! Sort of like a check, where you write the dollar amount with numerals and then write it with letters as well. It's actually quite amazing--with all the talk of voter confusion, butterfly ballots, and the like--that nobody realized this simple mistake would be so common. Nobody until now, that is.
The Lake County numbers contain another stunning surprise, in that they dramatically validate what might be called the "Sloppy Dem Thesis"--the folk wisdom that says Democratic voters (being less experienced, or less well-educated, or less anal, or whatever) tend to make ballot errors more often than Republican voters. Lake County, remember, is a Republican county that Bush carried by a wide margin. Yet the recoverable ballot errors (at least the overvote errors) ran heavily in favor of Gore. (Even other, more problematic ballots that the Sentinel didn't count--such as when a voter attempted to erase one mark--"fell heavily in Gore's favor.") If Gore picked up votes on a recount in Lake County, where wouldn't he pick up votes?
The writer Murray Sayle once joked that there are only three real stories in journalism: 1) "Arrow points to defective part;" 2) "We name the guilty man;" and 3) "Everything you thought you knew about this subject is wrong." The Lake County story comes close to qualifying for the third category. Consider its apparent implications:
Gore was mistaken: Gore went for hand recounts in four Democratic counties rather than a broad statewide recount. He's been criticized for grabbing at a quick political advantage instead of taking a gamble and doing the "right thing." But it's now clear that the right thing wouldn't have been much of a gamble for him at all. If the Sloppy Dem Thesis is as correct as it was in Lake County, Gore would have gained votes all over the state, in pro-Bush counties as well as Democratic counties.
Gore was doubly mistaken to focus, laserlike, on the undervotes, ignoring the potential harvest of uncounted votes in the overvotes that resulted from voters making the Write-In Error.
Gore was a total fool, in particular (and in hindsight), to ignore the massive overvote of 21,000 ballots in Duval County. According to Richard Cooper's post-mortem in the Los Angeles Times, Gore aides assumed these votes were unsalvageable. Only after the deadline for requesting a recount had passed did the Gore team meet with local allies and learn that many of the overvotes contained the Write-In Error--and might have been counted for Gore.
The press was equally wrong to follow Gore's lead and cover the recount as if the undervotes were the whole story.
Gore and the press also missed the boat by focusing almost exclusively on the voting problems in "punch-card" counties. Optical-scanning counties may have held large troves of votes, too. True, it's not clear how many optical-scanning counties decided, as Lake County did, to not examine overvotes for the Write-In Error. (At least two jurisdictions in fact counted such votes, according to the Orlando Sentinel.) The Sentinel is even now trying to find out how many other counties acted as Lake County acted. But, since Lake alone yielded 130 new net Gore votes, it would only take three or four similar counties (out of 38 using the optical-scanning system) to put Gore over the top.
The whole chad debate was unnecessary! By focusing on "punch-card" undervotes, Gore was inextricably drawn into the murky and morally ambiguous world of chad. He wound up throwing the full force of his advocacy behind the highly questionable Delahunt standard, under which merely "dimpled" chad can be counted as clearly intended votes. But if the Lake result is indicative, Gore didn't need dimpled chad! He would almost certainly have won a full statewide recount under the strictest chad standards--if, that is, the recount included the overvotes in the punch-card and optical counties. The whole debate over dimples was a needless drain of Gore's legal and moral resources.
James Carville was right! He boasted that Gore would win a recount even without dimpled chad.
Kausfiles was wrong to ridicule Carville for this boast.
James Baker was right--in a strategic, not moral, sense--to fight all manual recounts instead of seeking his own hand counts in pro-Bush counties. The Lake County result shows that even in Bush counties a hand recount would probably have helped elect Gore, thanks to all the Sloppy Dems.
Kausfiles was also wrong to suggest that Baker "blew it in Florida" with this "no recount" strategy. Please don't click here to see just how wrong.
Slate's "Ballot Box" was wrong, in hindsight, to estimate that Bush would win a statewide recount under strict standards but lose under loose standards. We now know Gore would probably have won under either standard. Jacob Weisberg made it clear that his Slatecalculation assumed that the Sloppy Dem Thesis was invalid (which it clearly didn't turn out to be in Lake) and that overvotes didn't matter (which it's now clear they do).
The Florida Supreme Court was wrong to order a statewide recount that was seemingly confined to undervotes. The overvotes should have been recounted, too (though if Gore had won with only the undervotes, then counting the overvotes would probably have only widened his margin of victory). But Lake County officials say they would have examined their overvotes if the recount had gone forward, and the circuit court judge presiding over the recount probably had authority to order that other overvotes be examined as well. So maybe the Florida recount would eventually have been fairer than it at first appeared.
But the Florida court was very clearly wrong, in retrospect, when it allowed a count of both undervotes and overvotes in mainly Democratic precincts in Miami-Dade County while adding to those results a count of only undervotes in mainly Republican precincts. The only way that apple-orange sandwich could be fair is if recounting overvotes never yielded anything. But we now know that overvotes can contain a huge (one in five) stash of salvageable ballots.
Finally, the exhausted post-concession press has been wrong in failing to give the Sentinel story the play it deserves. The New York Times' news pages, according to Nexis, have simply ignored the Lake County recount and its implications. (Only Maureen Dowd's op-ed column has mentioned it.) A recent NYT editorial defending the utility of press recounts didn't bother to note the Sentinel's effort, even though it helps make the editorial's point.
Let the orgy of recrimination continue!