Five years ago yesterday, "georgia10" posted a diary called "Armando's Challenge, Or The Informed Citizen's Guide To The 2004 Election." I’ll explain that title below the fold. The associated 57-page document was soon retitled "Eye on Ohio" (links below), and it set forth the evidence that election fraud in Ohio had given Bush his reelection. As the introduction explained:
If you are looking for a smoking gun here, you will not find it. Instead, you'll read about a series of complex and seemingly impossible events, from voter suppression, to vote tampering, to possible cover-ups. It is the totality of the circumstances that compel, at the very least, a full-blown Congressional investigation into the matter. Is this enough evidence to cast doubt over the election of George W. Bush? Was Ohio really a "blue state" in this election? As you read, you'll realize that the weight of the evidence strongly suggests such a conclusion.
At the time, "Eye on Ohio" offered an urgent argument that Congress should investigate the 2004 election before (instead of?) sealing Bush's victory. Now it can be read to consider where election integrity efforts stand five years later.
Brief introductory meta
I wasn't a Kossack in late 2004 and early 2005 -- I joined sometime in the spring of 2005, and quickly became a semi-willing combatant in what might be called the "fraudster wars," in particular criticizing arguments that the exit polls evinced massive vote miscounting. I am not interested in reenacting those wars here. We can certainly discuss substance, but I humbly beseech any and all readers not to hijack the diary with nasty comments about individuals or (perceived) groups -- or, for that matter, paeans to glorious martyrs. I hope you'll agree that we have better things to do.
Why "Armando's Challenge"?
For those who have forgotten or who never knew, Armando was a prominent Daily Kos front-pager, who now posts as "Big Tent Democrat" at TalkLeft. So, the point of "Armando's Challenge" was what merited exposure on the DKos front page. Back in mid-December 2004, Categorically Imperative commented:
Well, the pieces in Ohio don't fit together, and we've got no evidence that the election wasn't fraudulent. Actually, the exit polls coupled with the circumstantial evidence that's been developed point towards fraud, or at least some massive malfunction in the entire system. There's as much of a basis to discuss Ohio fraud as there is to discuss Bush being AWOL.
A few comments later, Armando responded (to rincewind) in part,
If Ohio is as solid as AWOL..., then I will write a front page post on it this weekend.
Post a diary. Ask that the best most solid Ohio evidence be demonstrated, and I promise, cross my heart, that if I am convinced that it is as solid as AWOL, I'll write it up on the front.
georgia10 and some others took up the challenge, producing "Eye on Ohio" (.doc version, 2.2 MB) (HTML version), which quickly received positive comments (ranging from respectful to effusive) from across the Kossack opinion spectrum.
Outline of the argument
It's risky to summarize a 57-page document with 15 sections. Let me hit a few high points:
"Eye on Ohio" opened with four definitions of fraud, starting with:
Fraud: n. the intentional use of deceit, a trick or some dishonest means to deprive another of his/her/its money, property or a legal right. (p. 4)
georgia10 added, "Ideally, the discussion here will demonstrate -- to some degree -- that there was an attempt by various individuals, as possibly coordinated attempts, to 'dishonestly deprive people of their legal rights.'" Note that while the introduction suggested that "the weight of the evidence" pointed to a Kerry victory, the argument for fraud in Ohio does not depend upon whether Kerry should have won. Also, while some observers construe "fraud" as referring specifically to attempts to rig the count (such as voting machine hacks), georgia10's definition also includes many forms of voter suppression. Failure to distinguish among these concepts has mired many a DKos discussion.
"Eye" points out that the voting machines are suspect. Diebold and ES&S, the two largest providers of Direct Recording Electronic (DRE) machines and optical scanners, have criminal connections and strong ties to the Republican Party; their equipment is known to be hackable. Triad, which (inter alia) provided the punch card hardware and software used in much of Ohio, also has strong Republican ties. There are some clear (and some less clear) examples of machine errors; "miraculously, almost all these 'glitches' favor Republicans" (p. 11).
"Eye" documents extensive voter suppression effort, beginning with purges of supposedly ineligible, or inactive, voters -- and private registration drives in which Democratic forms apparently were destroyed. It also notes that in (heavily Democratic) Cuyahoga County, Ohio, many registration forms appeared not to have been processed (p. 14). (In passing, it notes Ohio secretary of state Ken Blackwell's bizarre attempt to reject registration forms not printed on 80 lb. paper.) It describes efforts to limit both absentee and provisional votes in Ohio and elsewhere. It documents individual "challenges" to specific voters. It describes attempts to intimidate voters or deceive them about their eligibility to vote.
"Eye" discusses the misallocation of voting machines in Franklin County, Ohio. It correctly notes that heavily Democratic precincts generally were subject to much longer lines. (One estimate suggests that about 22,000 voters in Franklin County left because of long lines, of which about 16,000 were Kerry supporters.)
At about this point, georgia10 offers a general theory of sorts:
I submit that the strategy undertaken in Ohio was a bifurcated attempt to both pad the Republican totals in areas utilizing electronic voting and to prevent traditionally Democratic voters from casting their ballots in paper ballot precincts. (p. 24)
The basis for the first "fork" is unclear, but "Eye" has a lot to say about voter suppression in Democratic counties -- regardless of their voting method.
"Eye" notes that the "incidents" reported to the Election Incident Reporting System (EIRS) tended to be concentrated in heavily Democratic counties, in Ohio and elsewhere (p. 25 and elsewhere). (This result is hard to interpret: Democratic counties tend to have more voters, and the EIRS was publicized through election protection volunteers and media sources that themselves had an urban and Democratic skew. It seems very likely that Democratic voters experienced more "incidents" than Republican voters, but the EIRS doesn't provide strong evidence about the extent of the imbalance.)
A section on vote counting alleges evidence of absentee vote "inflation" in five Trumbull County communities; cites some odd preliminary results; claims that "[a] down-ticket Democrat polled 257,000 more votes than Kerry in Ohio" (p. 29 -- see below under "Glitches"); and cites the "caterpillar crawl" errors in Cuyahoga County, where hundreds of votes cast for Kerry in one precinct were counted as votes for other candidates (or undervotes) in other precincts at the same polling place. It cites reports that some punch card ballots had been prepunched for Bush. It also cites the higher spoilage (undervote and/or overvote) rates statewide in counties that used punch-card voting, and notes that in Cuyahoga County, at least, spoilage rates were highest in heavily African-American precincts.
A section on exit polls notes that Kerry led in the initial exit poll estimates nationwide and in several swing states, including Ohio, that he eventually lost. "Eyes" observes that these discrepancies are extremely unlikely to have been due to chance, and says that "[t]here has not been a scintilla of evidence to support the argument" that Bush voters participated in the exit poll at lower rates than Kerry voters (p. 34).
A section on the Ohio recount notes that dozens of counties either failed to conduct proper 3% random samples, or tampered in some way to avoid discrepancies that would trigger full manual recounts, or simply refused to conduct full recounts despite discrepancies. (To conceal massive fraud, or just to avoid full recounts?) "Eye" then summarizes attempts to investigate and to receive judicial relief, most of which had made little headway -- and never did make much.
georgia10 concludes in part:
119,000 votes. That is the margin of victory in Ohio. That number does not seem that large in light of how many thousands upon thousands were systematically removed from voter rolls, were unreasonably challenged, were intimidated, were told that Democrats vote on the 3rd...it's not such an insurmountable number when you think of all the testimony of "vote-hopping," of pre-punched cards, of long lines....
The warning alarm has reached a fever pitch; something is terribly wrong with our electoral system, and unless we do something immediately, there will not be any legitimacy left in our election. If we do not address these issues now, 2008 will bring more aggressive intimidation, more vulnerable software, more tried-and-true methods of shaving off or padding votes, more stonewalling and hiding the truth.
The next vote that will be stolen may be yours.
Now what are you going to do about it?
"Glitches"
"Eye" was written in haste, and some of the material probably would have been cut or substantially recast if time had permitted. I will mention a few points that have recurred in other sources.
P. 11: "In 2002, all three Republican candidates in Texas received the exact same number of votes: 18,181." This statement should refer to three candidates in Comal County out of 35 on the ballot -- not all Republican candidates in the county, much less in the state. (In the final returns, by the way, all three candidates received 18,183 votes.) "Eye" adds that "no one investigated the matter," but really it isn't clear what would be investigated. The county clerk reported that the vote counts were different in every precinct; the opposition vote totals vary, too. To believe that 18,181 evinces fraud, one has to believe that the vote-stealers deliberately tipped their hand -- and some people do believe that 18,181 means "AHAHA." This seems like a big stretch to me.
(I've written elsewhere on the Siegelman election in 2002, briefly discussed on p. 11. If contemporaneous news accounts are correct, that appears to have been a "glitch" favoring the Democrat, Siegelman, which was promptly corrected. However, in the absence of a recount, the facts will remain contested.)
The "down-ticket Democrat [who] polled 257,000 more votes than Kerry in Ohio" was C. Ellen Connally -- but she didn't actually do that. Kerry got over 2.7 million votes, Connally under 2.1 million. However, the assertion went that there were 37 counties in which Connally ran better in margin versus her opponent than Kerry ran versus Bush, and the sum of these differences was about 257,000 votes. Overall, Connally lost by over 284,000 votes -- more than twice as many as Kerry -- so she obviously did much worse than Kerry in the other 51 counties. Why is this suspicious? Actually, no one ever offered a good reason.
The exit poll discussion occupies six pages (out of 40 in the body of the report), to no good effect. Like similarly oriented discussions at the time and since, "Eye" never even mentions that many of the exit poll projections -- such as a double-digit margin for Kerry in New Hampshire -- were facially implausible. Any comparison of the projections to pre-election expectations provides far more than a "scintilla" of evidence that the exit polls were awry. "Eye" also missed the fact that the 1992 exit polls were almost as far off as the 2004 exit polls. As a matter of logic, that might evince massive fraud in both years -- but I haven't seen a serious case for fraud in those elections and not the intervening ones. I could go on. I've done it before. :)
Brief assessment
One can agree with georgia10 that 119,000 votes is "not such an insurmountable number," and still think that Bush probably won Ohio no matter how one looks at the question. That is what I think. With respect to vote miscounting, I see little evidence of it apart from a small number of Cuyahoga County precincts in which votes obviously were counted differently than cast (such as the one where Michael Badnarik got 164 votes). I concur with Walter Mebane and Michael Herron's conclusion that there is "strong evidence against the claim that widespread fraud systematically misallocated votes from Kerry to Bush" (p. 2). Assuming that is true, it still leaves the possibility of stuffing votes for Bush and/or stealing votes from Kerry -- but I don't see evidence that either of those happened on a massive scale, either. I don't see huge spikes in turnout or Bush vote in Republican counties, nor comparable evidence that large numbers of Kerry votes were destroyed. There were actually substantially fewer "spoiled" votes in Ohio in 2004 than in 2000. Despite gratuitously restrictive rules about accepting provisional ballots, all but about 35,000 were counted (and not all of those would have been Kerry votes). As for (deliberate and/or inadvertent) voter suppression, it can't be closely quantified, but the Democratic National Committee's Voting Experience Survey indicates that its net impact was probably substantially less than the margin. On close analysis, I think it's hard to get to 119,000 votes, at least by changing things that can be attributed to fraud or malice. (Yes, crazy terse, but this diary is already way too long.)
In January 2005, whether Kerry won Ohio of course was an urgent question -- but it was never the only important question. For a voter who was wrongly turned away at the polls, or intimidated while there, or who had to leave because of the lines, it is not much help to be told (rightly or wrongly!) that similar events probably did not alter the election outcome. Similarly, if your voting machine appeared to be "glitched," it isn't very reassuring to be told that it was probably an isolated instance -- and if it occurs to you that the vote count could be wrong without any visible sign, then you may just feel like hollering that there has to be some way to check the count.
Five years later
In 2004, it seemed for a while as if Direct Recording Electronic (DRE) voting machines without a voter-verifiable paper record (VVPR) might take over the country. (DREs sometimes are called "touchscreens," although not all of them actually use touchscreens.) According to statistics compiled by Election Data Services, DREs with and without VVPRs more than doubled their reach between 2000 and 2004 -- from counties with 12.4% of registered voters in 2000, to 29.2% in 2004 -- a growth rate almost three times as fast as that of optical scan equipment.
As verified voting activists spread the word about the many failings of DREs, the tide turned toward optical scanners. In my opinion, optical scanners are preferable to DREs for two big reasons. First, they use an auditable paper ballot that in most cases was hand-marked by the voter -- while even the best DREs generate a paper record that the voter may or may not actually verify. Second, the time to scan a ballot is much shorter than the time to enter one's votes into a DRE, making it cheaper and easier to avoid long lines. DRE's voter share further increased to 37.6% in 2006, but optical scanners spread more quickly in this two-year period. By 2008, the DRE share actually decreased to 32.6%; optical scanners were used in counties with over 56% of registered voters. The trend toward paper remains fragile, with widespread proposals to implement Internet voting (at least on a pilot basis) despite warnings that it can't presently be done securely. That said, I think the turn against DREs is a remarkable victory by any reasonable standard.
Optical scanners don't solve the vote count problem all by themselves. Scanners can intentionally or accidentally be configured to miscount arbitrarily many votes. To provide verifiability, the ballots have to be secured after the election, and some of them have to be audited to see if they were counted correctly. In order to confirm election outcomes, the audits should be designed to be risk-limiting (see definition in point "a") -- for instance, all else equal, closer elections should be audited more stringently. Risk-limiting audits have been piloted in California, but they are not yet routine anywhere. Some states have audits large enough to uncover evidence of outcome-altering miscounts in most cases, but most do not require further counting no matter how alarming the audit results may be. Moreover, provisions for securing the ballots after the election often are woefully lax. Some activists talk as if op-scan is almost useless (or worse than useless) in light of these limitations. I don't agree. To me, the Franken-Coleman recount dramatizes how, even in an extremely close election, paper ballots can yield a more trustworthy result than paperless DREs ever could. But we have work to do in every state.
As seriously as I take the security and reliability concerns about voting and tabulation machines, I believe that most of the problems and shenanigans in 2004 had to do with voter suppression or, more generally, barriers to voting. A good starting point for assessing progress and setbacks since 2004 is Advancement Project's report released last month, "Barriers to the Ballot." The report notes,
Because the vote difference [in 2008] was large enough, no official microscope will be applied to the administrative failures during the heat of the election and its immediate aftermath.... [T]he 2008 results –- and, indeed, the improved performance of many states –- have led many to breathe a sigh of relief.
But... serious problems continue to undermine the infrastructure of American democracy. Indeed, nonpartisan political scientists have already estimated that approximately eight million registered non-voters did not vote because of administrative problems -– such as long lines at the polls, registration issues, and absentee ballot issues. Of those, 3 million were registered voters who could not vote because of registration problems. This is not indicative of a system that can effectively perform and engender confidence in a close election. [page 3, footnotes omitted]
Because elections are locally administered and chronically underfunded, the battle to protect people's right to vote would be hard and complex even if no one actively opposed the cause. But it isn't futile! The Advancement Project report shows how extensive preparatory work by voter protection organizations helped to forestall many problems. Such efforts will have to continue even if the threat of massive vote miscount is substantially mitigated. Perhaps this is what (representative) democracy looks like.
Summing up
In a 2006 paper on the exit poll controversy, I quoted Walter Lippmann: "...when there is panic in the air, with one crisis tripping over the heels of another, actual dangers mixed with imaginary scares, there is no chance at all for the constructive use of reason." Like many Kossacks, I have thought that many arguments about fraud in the 2004 election harmfully mixed actual dangers with imaginary scares. I don't think that "Eye on Ohio" entirely avoided that pitfall. Nevertheless, georgia10 and her collaborators set a new standard for disciplined discussion of election integrity issues here on Daily Kos. Bitter that [names redacted] had dominated the fraud debate? Here was a copiously documented 17,000-word survey that didn't mention either one. Annoyed by an obsessive focus on voting machine issues? Here was an attempt to piece together the big picture. Burned out on Breaking Bull$#!t Du Jour? Here was a serious effort to winnow the wheat from the chaff.
Now what? The great thing about such a sprawling issue is that there is lots of different stuff to do. I happen to spend crazy amounts of time working on post-election audits, partly in the hope of avoiding future exit poll arguments. But that is one piece among many -- the specifics vary from place to place -- so whatever you enjoy, you can find a role. Of course, if you're hard-core enough to make it to the end of this diary, I bet you already have.