In the early 1980s, bookstores were featuring a book clearly aimed at baby boomers like me who were just then entering our thirties. Life Extension: A Practical Scientific Approach promised its readers could live for a healthy 100 years. Knowing basic mortality statistics, I checked to see whether the authors had addressed the leading cause of premature death for my age group. I looked in the index for 'seat belts' and found nothing. I looked for any mention of drunk driving–still socially acceptable at that time. Nothing.
How scientific could the book be if it neither mentioned the major risk of premature death nor described the easy practical steps that could reduce it? That book came to mind as I read the recently released Legislative Audit Bureau evaluation of the Government Accountability Board.
While never mentioning the shamefully unanswered question of whether our election results are accurate, the report lavishes attention on risks that—at their worst—pose only remote threats to our elections’ outcomes.
- The report discusses (2.5 pages) votes cast by dead people (243 people died after submitting an absentee ballot but before Election Day—how dare they!) and demands the GAB report to the Legislature about how it is going to reduce the average time it currently takes to invalidate a deceased voter’s registration (a whole 54.9 days after the voter's death);
- The report discusses (4.5 pages) the issue of felons who vote (possibly as many as 33 statewide in 16 elections over the past four years) and demands the GAB devise procedures for more promptly identifying and prosecuting those voters;
- The report discusses (2 pages) the issue of verifying information that citizens provide upon registration. Out of 531,407 registrations, municipal clerks provided no evidence they had verified the information for fewer than 0.1%. Without pointing out the infinitesimal likelihood that these 432 unverified individuals represent a conspiracy to steal elections rather than predictable clerical error, the LAB recommends GAB fix the problem and report back to the Legislature.
But what about our legally cast votes? Are they counted accurately? Silence.
Not one of the report’s 92 pages addresses the fact that although voting-machine output is occasionally found to be erroneous, GAB’s policies ensure that Wisconsin routinely uses it without verification as our final election results.
In discussing training of municipal clerks (4.5 pages), not one paragraph addresses the fact that GAB’s training includes nothing to alert clerks to the need to be alert for electronic miscounts or instructions on how to monitor voting-machine accuracy.
Not one paragraph addresses the fact that GAB’s instructions for pre-election voting-machine tests allow clerks to use ballots marked with the same number of votes for each candidate, so that even if the machine is counting Jones’ votes for Smith and Smith’s for Jones, the test will not detect it. Auditors made no apparent effort to determine whether clerks even conducted the tests, and failed to note that GAB never checks either.
The Bureau continues its failure to see the election-results forest for the individual-vote trees even in its two-page discussion of post-election audits. That discussion is devoted to the very few ambiguously marked ballots—those on which humans can discern voter intent that machines cannot read—and how they should be handled in post-election audits. The Bureau ignores the major problems with GAB practice and state law, such as the problem that only one election every two years is subject to verification at all, and that GAB encourages clerks to delay that verification until after election results are certified—a practice that eliminates any value the audits might have to protect our final election results from the effects of electronic miscounts.
This lack of interest in accurate election results must be considered against the background of recent elections. In November, voting machines were programmed in Dane County to miscount a referendum, a mistake that was caught only because the error caused most ballots to be read as blank. It was only luck that the error ignored the votes instead of counting no-for-yes and yes-for-no). Had the votes been flipped, the false results would have been outside the recount-trigger margin and would have been declared final without verification or correction. Earlier this year, ballots were destroyed in Green County before they could be recounted, an error that would be highly unlikely if municipal clerks were required routinely to safeguard ballots for post-election audits. Nearly every recount has discovered some errors, yet no one routinely audits races that are not recounted, so no one knows how many candidates have been sworn in to office on the basis of a miscount.
Finally, no one investigates or even collects data about the miscounts discovered by accident—the Stoughton referendum miscount will soon fade from official memory with no procedural change to prevent or detect the next such miscount, just as the Medford miscount did.
This irrationality is damning evidence that our legislators are laser-focused on partisan concerns—Will all my party’s supporters be able to vote? Is there a way to make it hard for some of the other parties’ voters?-- while any statesmanlike care for the overall integrity of our elections gathers dust in the legislature’s archival storage.
Are citizens, too, interested only in those election-integrity issues that are likely to benefit their side? Or will citizens engage their local election officials in the effort to detect and correct miscounts before they deprive us of our right to self-government? It's up to us.