Monday, November 28, 2016

Recount! Why, and how the numbers look today

Looks like the November election isn’t quite done in Wisconsin yet. In addition to the recount in Western Wisconsin to verify that Senate Dem leader Jen Shilling beat Dan Kapanke, we now have a statewide recount for the tight presidential race, courtesy of Green Party candidate Jill Stein and Independent Rocky de la Fuente. The notice was filed on Black Friday and procedures for the recount were OK’d by the Wisconsin Elections Commission today, assuming Stein and/or de la Fuente put up enough money to pay for it.

You can read the Stein campaign’s recount filing by clicking right here. It gives a brief explanation for the recount, and attaches several articles from the past few months detailing hacks into the Democratic National Committee and Clinton campaign’s emails, including an October warning from the Department of Homeland Security which called the hacks “consistent with the methods and motivations of Russian-directed efforts.” At that time, the US DHS explicitly said Russia has messed with elections in other countries, and was likely trying to do so in the United States.
These thefts and disclosures are intended to interfere with the US election process. Such activity is not new to Moscow—the Russians have used similar tactics and techniques across Europe and Eurasia, for example, to influence public opinion there. We believe, based on the scope and sensitivity of these efforts, that only Russia's senior-most officials could have authorized these activities.

Some states have also recently seen scanning and probing of their election-related systems, which in most cases originated from servers operated by a Russian company. However, we are not now in a position to attribute this activity to the Russian Government. The USIC and the Department of Homeland Security (DHS) assess that it would be extremely difficult for someone, including a nation-state actor, to alter actual ballot counts or election results by cyber attack or intrusion. This assessment is based on the decentralized nature of our election system in this country and the number of protections state and local election officials have in place. States ensure that voting machines are not connected to the Internet, and there are numerous checks and balances as well as extensive oversight at multiple levels built into our election process.
Of course, just because it would be “extremely difficult” to hack an election, it doesn’t mean that it is impossible. And even with the decentralization that the US DHS mentions (which is especially true in Wisconsin, as votes are generally reported by township and city/village level), that doesn’t mean someone couldn’t mess with a few totals or have a few pre-made ballots ready to stuff, especially if they knew the right way to sneak it into certain places throughout the state.

Also part of Stein’s filing is testimony from J. Alex Halderman, a University of Michigan Computer Science/Engineering Professor. He explains how hacking an election could occur, why Homeland Security’s confidence in previously reported tests may be misguided, and why we need a recount to make sure these elections are on the level.
13. One explanation for the results of the 2016 presidential election is that cyberattacks influenced the result. This explanation is plausible, in light of other known cyberattacks intended to affect the outcome of the election; the profound vulnerability of American voting machines to cyberattack, and the fact that a skilled attacker would leave no outwardly visible evidence of an attack other than an unexpected result.

14. The only way to determine whether a cyberattack affected the outcome of the 2016 presidential election is to examine the available physical evidence- that is to count the paper ballots and and paper audit trail records, and review the voting equipment, to ensure that the votes cast by actual voters match the results determined by the computers. For ballots cast through optical scanners, a manual recount of the paper ballots, without relying on the electronic equipment, must occur. Using the electronic equipment to conduct the recount, even after first evaluating the machine through a test deck, in insufficient. Attackers intending to commit a successful cyberattack could, and likely would, create a method to undermine any pre-tests. For votes cast on electronic voting machines, the paper audit trail records must be counted, since the electronic records stored in the machines could have been manipulated in an attack. Voting equipment that might yield forensic evidence of an attack includes the voting machines, removable media, and election management system computers. Paper ballots, paper audit trails, and voting equipment will only be examined in this manner if there is a recount.

15. A recount is the best way, and indeed the only way, to ensure public confidence that the results are accurate, authentic, and untainted by interference. It will also set a precedent that may provide an important deterrent against cyberattacks on future elections.
To be clear, Halderman isn’t saying that the election was necessarily rigged, but that there is a lot of smoke arising from related incidents and hacking, and that a recount could clear some of that smoke away.

The Wisconsin Elections Commission did take one step today that went against the wishes of Dr. Stein and Prof. Halderman, as it rejected their call to have the entire recount be done by hand, instead leaving the “hand-count vs machine” decision up to the individual counties. Stein has responded by filing suit in Dane County to order the recount to be done by hand, but for now, it looks like counties will make their call in the next couple of days and get to work later this week.

The last link I’ll forward to you on the recount is the Wisconsin Elections Commission’s release of the uncertified statewide results for president in Wisconsin. As you’ll see, the gap between Trump and Clinton has shrunk by nearly 5,000 votes from the 27,000 difference that was reported on Election Night, and a higher number of reported votes lowered both of the top-two candidates’ percentage share.

Election Night vs Elections Commission, Wisconsin
Election Night
Trump 1,409,467 (47.9%)
Clinton 1,382,210 (46.9%)

Elections Commission
Trump 1,404,000 (47.2%)
Clinton 1,381,823 (46.4%)
DIFFERENCE 22,177 (-5,080)

Remarkably, these totals would mean Trump got fewer votes in Wisconsin than Mitt Romney got in 2012…when Romney lost by nearly 7% to Barack Obama. This is because of a much higher share of 3rd Party and write-in votes than we had 4 years ago (6.4% in 2016 vs 1.3% in 2012), and because the total number of votes in Wisconsin dropped by more than 93,000 vs four years ago. That’s quite an adjustment, and further lends me to lean towards wanting verification that this is legit.

It also seems telling that our (alleged) President-elect and Governor Walker and other GOPs are now squawking about the fact that this recount will likely go on, with Walker taking the extra step of lying about taxpayers having to pay for the recount (they don’t, it’s being paid for by the Stein and Rocky de la Fuente campaigns). Methinks the slimy righties are protesting a wee bit too much here, which they tend to do when there’s a danger of something being revealed that they don’t want people to know. Yes, the last 6 years in Fitzwalkerstan have left me that cynical and suspicious of this crew, and it makes me all the more interested to see what might be found in Wisconsin over these next 2 weeks.


  1. UPDATE- Now the Wisconsin Elections Commission claims the recount won't cost $1.3 million, but $3.5 million!

    Here's the spreadsheet from the Elections Commission as to how they arrived at that number. I have no idea how those figures were arrived at (it doesn't go deeper than the topline numbers), but it sure smells of BS. Apparently if it's less than that, the money will be refunded, but seems like a lot of people really don't want it to happen, do they?

    Makes me want to give a little bit to the recount effort to find out why.

    1. Sounds like SOP to me. Our right wing has a history of overstating the cost of things they don't want to do.

      On another topic, I can say from experience that government offices of all sorts get port scanned on a very regular basis, and that's been going on for years. Malicious software infection is a pretty regular occurrence.

      I'm not a security expert, so I can only speak to a certain level of this, but the notion that a dozen or more county level computers could be infected in the way Halderman suggests is absolutely believable.

  2. How about having 2 open boxes of absentee ballots on tables in full view and reach of voting public.

    Aren't ballots supposed to be kept in a secure location until voting place closes?

  3. This comment has been removed by a blog administrator.

  4. Will this recount look at Feingold and idiot Johnson's race? Would help to have another Dem in the Senate to oppose Trump.

  5. What is the cost per counted vote deviation from county to county

    1. You think I'm doing that? :P My real job doesn't allow time for it now, but both the vote spreadsheets and alleged county cost of the recount are linked on this post, so tell me what you find

    2. Some numbers here. Very, very wide range of costs from different counties.

  6. I don't expect a recount to change the vote by anything close to 20K unless there was severe election fraud in play.

    Going out on limb here, but it seems like we don't need a hand-recount of the entire state as much as we need a targeted audit to determine whether there were dirty tricks in play.

    I don't know the math behind it, but I would think a good strong audit of maybe 10-15% of the precincts would be sufficient to pick up interference at the level necessary to turn the election.

  7. In response to Anonymous last night (11:24 pm, 11/29): the spread in cost per vote was from 15.6 cents per ballot (Menominee County) to $8.54 (Oneida County). I understand that Oneida has since changed its estimate so the highest cost county is now Pierce, at $6.68 per vote. The average estimated cost per vote is $1.32. Dane ($1.12) and Milwaukee ($1.23) both brought the average down. Thank goodness for blue counties!