Michigan recount hits a snag
And it’s quite a large snag:
One-third of precincts in Wayne County could be disqualified from an unprecedented statewide recount of presidential election results because of problems with ballots.
Michigan’s largest county voted overwhelmingly for Democratic candidate Hillary Clinton, but officials couldn’t reconcile vote totals for 610 of 1,680 precincts during a countywide canvass of vote results late last month.
Most of those are in heavily Democratic Detroit, where the number of ballots in precinct poll books did not match those of voting machine printout reports in 59 percent of precincts, 392 of 662.
According to state law, precincts whose poll books don’t match with ballots can’t be recounted. If that happens, original election results stand.
“It’s not good,” conceded Daniel Baxter, elections director for the city of Detroit.
He blamed the discrepancies on the city’s decade-old voting machines, saying 87 optical scanners broke on Election Day. Many jammed when voters fed ballots into scanners, which can result in erroneous vote counts if ballots are inserted multiple times. Poll workers are supposed to adjust counters to reflect a single vote but in many cases failed to do so, causing the discrepancies, Baxter said.
Even so, Baxter said it’s unlikely all 392 of the city’s precincts with mismatched numbers will be disqualified from a recount. The city is in contact with elections officials at the state of Michigan and Baxter predicted the numbers will match when the ballot boxes are re-opened for the recount, which starts Tuesday in Wayne County at Cobo Center.
It took me a moment to figure out what this was actually saying, because I’m unfamiliar with the way votes are counted in Michigan. It appears the voters fill out paper ballots, which are then fed into optical scanners to be counted. They are alleging that it’s the latter process that went awry, as well as human error by poll workers (harried? purposeful?) who failed to adjust for the broken scanners.
So were ballots counted multiple times that were “inserted multiple times”? Inquiring minds want to know. And if poll workers see that many scanners are broken, shouldn’t the ballots be hand-counted right off the bat, and the scanners ditched?
The paper ballots do still exist in Michigan, though, and they could all be counted in the recount. But the law states that they should not be counted if this discrepancy exists—and boy, does it ever exist. To me, it also seems logical to suspect that, if they all were to be recounted, Hillary Clinton’s totals would actually drop, because the multiple recording of votes in these heavily Democratic districts would be corrected.
[NOTE: Many of you probably remember horrors of the Minnesota recount of 2008 that put Al Franken in the Senate. But that was a much closer contest to begin with than the one in 2016 in Michigan, in which Trump is currently ahead by close to 11,000 votes statewide. In the Franken case, the constantly-changing difference between the two candidates always numbered in the hundreds, and sometimes even the low hundreds. With that sort of margin, it’s much easier to pull some shenanigans, although the “ballots ‘found’ in a car” situation apparently wasn’t as bad as legend has had it.]
It’s hard to mess with the Michigan system unless you have a way to “rix” the count by hacking into the machine. I don’t think the machines are connected to the web, so it’s not going to be an outside job.
The ballot is somewhat larger than a sheet of paper from a legal pad, stiffish, almost carboard. Counting two ends and two sides, there are four ways to insert it into the counter. It’s certainly possible that somebody isn’t watching or doesn’t get the DIRECTIONS on the ballot and the machine. But eventually, you’ll get it right. But perhaps somebody puts it in part way, pulls it out and turns it over….something like that. At what point in the supposed process there is a “count” is a question. Poll workers aren’t watching you do this, so an immediate correction isn’t in the cards.
Looking at the news story, it seems most of them were off by only a few ballots
“n Detroit, 158 of the 392 precincts with ballot discrepancies had just one extra ballot accounted for either in the poll book or in the ballot box, according to the Wayne County’s canvassing report.
For suburban Wayne County, 72 percent of the 218 precincts boxes with discrepancies in the number of ballots were off by one ballot.
The other ballot discrepancies in Detroit and Wayne County precincts ranged between two and five ballots, according to the report.”
I read the same story earlier this morning and my reaction is much the same, though the story itself omits some important details that a reasonably bright person would want to know, and I count someone with an average IQ as reasonably bright. So, if any readers here are election officials in Michigan, or know one personally, please ask them for clarification. In any case, below is my interpretation of the story, with added suspicions about what it means:
Michigan voters fill out paper ballots that are then fed by an official into an optical scanner that both counts the votes for all candidates in all races, and counts to total number of ballots fed through machine. Due to people not voting at all in a particular race, it is perfectly reasonable for the total number of counted ballots to not match the summed totals of candidates unless there is a separate count for “No vote” in this race”, which I will assume there has to be (anyone know?)
So, when the ballots from a machine/s are combined, bagged, and sealed, are the total ballots in the bag hand-counted, or do you enter the number what the machine/s gave you? Of course, the vote totals for the listed candidates totals are the machine/s counts. So my interpretation is that the numbers written on the bags are different than what was reported for that bag on election night.
The above sounds odd to me. Let’s suppose the Wayne County official was truthful about this being an error caused by repeated feeding of ballots through balky optical scanners without resetting the counter itself. Wouldn’t all the counts be offset the same way? Why would this lead the wrong number being entered on the bag that was actually different from vote count on November 8th and 9th?
What I am asking is- what exact numbers on the bags are entered in the first place? If a collection of ballots was machine counted before being bagged, shouldn’t the numbers provided on election night be the same as those entered on the bag itself? I got the impression from the story that the bags are opened and initially hand counted for sheer number just to make sure the total number of ballots corresponds to what was entered on the bag the day of or after the election. If, after two counts, this number does not agree with what the bag showed, it can’t be officially recounted for candidate totals because it is a priori evidence that ballots were removed or added to the bag.
The Wayne County official’s explanation can be correct and innocent at the same time. I can certainly imagine accidental mistakes of this type being made, but I can also imagine nefarious reasons for these counts to be off in this way, and it is quite interesting that these mistakes seem to be predominantly occurring in Detroit, Flint, and Lansing. At this point, I want to see a hand recount in Wisconsin and Michigan. My suspicion is you will find that ballots were fed through the counting machines multiple times in different locations, and on much larger scales than statistically predicted for simple human error.
Whycoh,
Some questions. My impression is that what you are talking about are discrepancies in numbers entered on the bag when it was sealed and the poll book numbers to account for the ballots used. These numbers would have been generated the night of the election or the day or two afterwards as the canvassers finished their work. A good canvasser is supposed to ensure these 1 or 2 ballots errors are dispensed with before the bag is sealed in the first place, and the Detroit News story seems to indicate that some canvass boards might have been lax in certifying results before accounting for these small discrepancies. If this is all that is at issue, I would agree that it is a minor one and the precinct should be recountable regardless.
However, that isn’t the impression I got from the first story. It sounded like to me that the bags have already been opened and the ballots inside are counted for sheer number to make sure the number entered on the bag is the same as hand counted, and this is done twice (or even more) if there is a disagreement. I would imagine if you had a bag of 1000- 10000 ballots, a hand count might make an error of one or two the first time, but a second count is likely to be more thoroughly done to ensure it is as completely accurate as is possible- you triple check stacks of 10 and triple check the numbers of stacks. It sounded like to me there are significant discrepancies in the numbers entered on the bag versus was was found when they were opened this week.
Yeah, well, “Detroit”. What else is new.
Blast from the past.
http://www.crainsdetroit.com/article/20110822/BLOG097/110829985/detroit-has-more-registered-voters-than-residents-over-18-census
And this, from 2005 Det News, is probably gone, but I add it for flavor,
Sunday, October 30, 2005 Detroit News exclusive report
Absentee ballots tainted?
“
Where I live in Florida, we use similar tech, and we’ve not had a problem. Not even in 2000, when we recounted 3 times (?) and got the same count every time.
You can find sample ballots here: http://www.leonvotes.org/Sample-Ballots/Sample-Ballots
Interesting that Yancey says the poll worker feeds the machine. Here, the voter is required to feed the machine. That eliminates a possible shenanigan, but the value added there is that the scanner will evaluate your ballot for over and under votes.
If rejected, it spits the ballot back out, and you can fix it or request a new ballot if you made an error. I want to say that there’s a “this end up” mark on the ballot so you know which way to feed the machine, but I just don’t remember that detail.
In any case, the counter does not count up until after the ballot has been scanned and accepted. The count is the count, and if the indicator says X ballots cast, the interior should also contain X ballots. And if you hand count the ballots, the computer tally and the hand tally should be the same.
This will sound horrible, but this ain’t rocket science. If Wayne County is having inconsistent tallies, there is something wrong.
When my mom worked the polls with such ballots she complained that many people didn’t quite get that you’re supposed to fill in the circle, not mark it with an X or a check mark. But that should be caught as an under vote.
Just for fun and historical perspective – while acknowledging that Curry is dead and with Castro now, and that absentee ballots are not the issue – I will post up another portion of the article I cited wherein it begins to offer up bullet points.
You will howl with laughter.
One thing I didn’t consider, tho, are write ins. But that should still tally a vote for the particular race even if it isn’t obvious whom got the vote.
So you’ll have something like
votes for all registered parties on the ballot + votes for write in = total votes cast on that machine
Here’s a video from the Michigan SoS office https://youtu.be/aizel-SzoBw?list=PLJHaXzzP47l_Ijy0GomiScH6pyu2FSub2
I tried to look for sample ballots from Detroit, but I got nothing.
We use the same sort of readers in our county.
Scenario one: Ballot count: 1000; machine total: 1001. The most likely cause is a failure to reset after a jammed ballot.
Scenario two: Ballot count: 1001; machine total: 1000. Not so easy to come up with a reasonable cause.
HOWEVER: in both scenarios, the paper ballots will not be recounted, and the original tally will stand. In Wayne county, that means that many precincts that were reported overwhelmingly for Clinton will not be subject to a recount adjustment: there will be no possibility of lowering her totals. Meanwhile, those precincts elsewhere, many of which went for Trump, are subject to recount, with the possibility of votes being shifted from Trump to Clinton.
I think that the “HOWEVER” is the key item of interest here. Recounting ballots only in areas where she lost has the possibility of benefiting Clinton, whilst locking in a large number of her votes. That is the tactic here–with the added effect of avoiding looking at discrepancies in the Wayne county ballots.
The recount advocates known there is no way to reverse the outcome. Its all about wanting to make it difficult to certify the election results and make djt’s win look like gwb’s 2000 victory: tainted
CBI,
I was thinking about this earlier when reading Whycoh’s comment about most of the errors being extremely small 1 and 2 ballot differences. While minor, and possibly due to exactly what IRA Darth described, and thus innocent, it is also quite possible that if one had done something nefarious, like simply transfer votes from one candidate to another in the reported count, one could try to cover this by simply intentionally making a 1 or 2 ballot error when sealing the ballot boxes/bags thus ensuring they never get a hand recount. This is why I want the recount now. Something doesn’t quite sound right here.
Of course, totally honest mistake in Detroit. The same type of error will be found in Chicago, Philadelphia, Baltimore, Milwaukee, etc., etc., etc.
The cause, very simple — those zombie voters keep stumbling around and bumping into things.
s the late, great, Richard Daley is reputed to have said about a particularly close and unexpected election result, “Well, we count the votes.”
Many of you are ‘looking at the squirrel’. This has nothing to do with recounts or counting procedures. It has nothing to do with shifting electoral votes to hrc or stein. Chase your tails, but catching your tail is pointless. My cats do it better.
We use the optical scanners too here in Suburban Philadelphia. It is like old SATs, fill in the circles. You feed the ballot into the scanner, and it accepts or rejects it. I cannot see how it would be counted more than once. You even get a receipt. The ballot is not returned if it is accepted, so it would be hard to scan more than once.
The premise of miscounts seems very unlikely unless there was actual fraud.
Maybe this is all obvious, but here is the deal with the optical scanner where I vote:
The machine takes the ballot and doesn’t give it back when the vote is tallied. If there is a problem, the scanner spits out the paper ballot and does not register a vote.
A poll worker stands at the scanner, instructing each voter- we feed our own ballot, which can be inserted in any direction. When the ballot is tallied, we are asked to verify that our vote was counted and we get one verification receipt.
We carry our marked ballot hidden in a folder until we feed it into the scanner. But, the sharpie markers bleed thru the paper, so a poll watcher glancing down, can actually see the dot pattern and make an accurate guess, even if the ballot is put in face down. I live in a small town and know everyone at the polling place pretty well, and they’re all dedicated and serious. Even so, I wonder about that. We’re lunatic about politics and everything here, just like everywhere.
If the Michigan scanners were spitting out the paper ballots and recounting the votes each time the ballot was reinserted, shouldn’t that have been raised as a problem during the election?
We all sign in to vote, and whether we vote or not appears to be public record– I certainly get a lot of pointed political email about my voting record, which freaks me out.
I don’t know what the procedure would be in my town if the machines became defective during the voting.
Sometimes people abstain from voting on the ballot to make a point. But, wouldn’t it be really obvious if the total votes cast for the candidates or propositions added up to a greater number than the number of people who signed in to vote that day?