Thread by @chrisd9r

I took today off to play Cyberpunk, but now that Trump's amplified the ASOG Antrim report, I guess I'll just start another thread to loosely collect my thoughts on the rest of it. You can find part 1 here:

https://twitter.com/get_innocuous/status/1338576587780812800

Let's go top-down. I'm working off of if you'd like to follow along. I can't guarantee that this thread will be accurate, coherent, or typo free because I've already started drinking Fernet, but I'll do my best.depernolaw.com/uploads/2/7/0/…

Quick note on my motivations: I'm a career infosec engineer. I fully acknowledge that electronic voting systems have a checkered history. But there has been zero evidence presented so far that clearly indicates this election was tampered with at scale.

The links drawn so far between Dominion, Scytl, Smartmatic, etc have been specious at best. And people with influence have been parroting these claims that their followers are eating up hook, line and sinker.

This 'evidence', all based in shoddy infosec snake oil and peddled by supposed "cyber terrorism experts", has lead to an increase of threats of violence and death aimed at elected officials, which is absolutely disgusting.

But infosec is complex, and as a result, it's easy for people to prey on people's lack of understanding. My goal is to do anything I can to help people see the flaws in the 'evidence' that's being presented.

On the flip side, if someone can present actual evidence tampering occurred, I will absolutely acknowledge it. Until then, I'm going to work with what's been presented as fact so far. So with all that out of the way, let's dive in.

Anytime someone says "dark and deep web" that's a big red flag that they have no clue what they're talking about.

The adjudication process is an election fraud greatest hit at this point. Adjudication is the process by which ballots that cannot be read by the machine are evaluated, so that the votes can be corrected and tabulated. The exact process differs between states and counties.

No one has been able to prove that the machines have been hacked or compromised, so most voter fraud claims have hinged on the adjudication process either switching votes or discarding them en masse.

Adjudication is a necessary feature of electronic voting. If the scanner messes up, or someone marks an X instead of filling the bubble in, the machine can't read the ballot. So adjudicators (typically at least 2, from opposite parties) are supposed to review and correct.

ASOG + co are alleging that machines have been intentionally configured to misread ballots so that they're marked for adjudication, and that the adjudicators are then throwing the ballots away or switching them to Biden. They try, and then fail, to back this up in the doc.

They present a table that shows the results as tabulated on 3 different dates. The 11/5 results are incorrect because Antrim failed to properly update the system to handle a new ballot. This is explained here: but ASOG disagrees on the cause. michigan.gov/documents/sos/…

ASOG continues to allege fraud while completely glossing over the fact that the re-tabulated results correctly gave Trump the votes he received and ended up with him winning the county. If there was fraud due to unnecessary adjudication, it seems to have helped Trump

A few things here:

1) I have no idea where the FEC .0008% is coming from

2) What are these 'events'? Why are there only 15,676 of them, if there were 20,082 votes cast? If it's not 1 event per ballot, what is the significance of the supposed 68% error rate?

For #9 - sure, I'll bite. But we don't know what these errors are, so we can't really judge their significance. I'll dive into how this number could be misleading in a bit.

They don't clarify what they 'reversed' means. It doesn't seem to mean that the votes themselves were switched, so I *think* they mean the paper ballots literally could not be read by the scanners and were 'reversed' out. Could happen if the ballot was deformed after being mailed

They have a sound point here. I haven't confirmed, but if there isn't an audit log of the adjudication activity, that's not great. But based off the other inaccuracies in the report, I'm hesitant to take that claim at face value.

Lot's of alleging of intent, but the doc lacks the evidence to back it up (unless it's in the redacted portions). It also ignores that there was a pandemic this year, and many ballots were mailed in, which could result in deformations that make them difficult to scan in.

Citing a tweet - DRINK!

If true, this should be verified by not-ASOG and looked into. Auditability is key. If we don't have this information, we can't verify ASOG's claims, which gives them the slack to fill in the blanks, which is exactly what they end up doing.

#16) Also strange if true, but itself not proof of anything.

#17) Demonstrably false, see the first tweet in this thread.

They MI SOS debunked this. The ballot reader config didn't match the ballots. It was updated, and the counts normalized. They allege intent without proving it. michigan.gov/documents/sos/…

Breaking news: when you connect a computer to the internet, it's connected to the internet. They don't prove that it *was connected during the election*, just that, I guess, they can connect it to the internet?

No, you don't have to deduce that, but go off.

Research is ongoing. however, based on the preliminary results, we conclude that the errors in this report are so significant that they call into question the integrity and legitimacy of this report. This casts doubt on the integrity of ASOG and their investigators.

ASOG really shows their hand in items 23/24. Too long to screenshot, but it's the standard Qanon talking points: Trump's going to invoke his 2018 executive order because of foreign interference in an election (that's not proven in this report). Also HUGO CHAVEZ!

I love how they sneak in that non-sequitur at the bottom that tries to connect Smartmatic, Dominion, Scytl and Venezuela. Again, more Qanon-sense.

ASOG visited Central Lake township and reviewed the voter roles. They noticed differences between the tabulation on 11/3 and 11/6. This, again, is explained by voting config not being correct, but they just refuse to accept that I guess freep.com/story/news/pol…

They waste a lot of space beating the same dead horse. The misconfiguration caused a significant vote discrepancy. It was corrected. That's it. That's the story.

If ASOG knew what they were doing, they could diff the 11/3 and 11/6 election definition's to see what changed between the two, and either verify or disprove the fact-check issued by the MI secretary of state. But they don't, so they didn't.

Moving from Central Lake Township to Mancelona, the only thing they can point out is that a test ballot process did not occur. Deep state must have skipped this one ¯\_(ツ)_/¯

Fun fact: one of the tools they use, Cellebrite, has a history of being sold to repressive regimes, which in turn used the data gleaned with the tool to do all sorts of horrendous things. Neat! vice.com/en/article/aek…

The next section basically sums up how the EMS servers were out of data. For computers that aren't supposed to be internet connected, this is not shocking.

Again, they do make a few good points. The EMS should have Bitlocker enabled, and people, for the love of god, don't share a password to a single login, especially one with super user privileges.

OK! Now on PAGE 17 we get to the meat of their claims. The errors! So many errors! How can you election with SO MANY ERRORS?!

Let's dive into this first claim: the 'election logs' for Antrim had 15,676 events, 10,667 of which were critical ERRORS *or* WARNINGS. They weasel that in there, but the distinction between an error and warning in programming is important, especially in this context.

All operating systems and programming languages let you log events at certain 'levels'. Those levels typically treat 'WARNING' and 'ERROR' differently. It's normal to see significantly more WARNING messages than ERRORs. docs.microsoft.com/en-us/previous…

You can't combine the two into a total and treat them both equally. If a developer wanted to log an error, they would log it as an error, not as a warning. ASOG does not break down how many of those 10,667 "errors/warnings" were warnings vs errors, so that number is worthless

So here they state that 1,222 out of 1,491 total ballots cast were reversed, and in some cases it was because the ballot was too large. But they leave out a ton of detail. Was this all in one township? Or Aggregate across the county? What does "reversed" mean in this context?

I also haven't been able to find any Dominion docs that break down the "size exceeds maximum expected ballot size" message. Without more details, it's impossible to verify their claims here.

Here, they simply state "A high "error rate" in the election software..."reflects an algorithm used that will weight one candidate greater than another", and that "In the logs we identified that the RCV or Ranked Choice Voting Algorithm was enabled"

To back up their claims, they don't offer the log messages that indicate RCV was enabled, but instead a....screenshot from a Dominion manual. So again, they make a claim, assert that it's correct, and then do not offer any actual evidence that can be used to verify.

Next, they mention that the software configured to divert write-in ballots for adjudication. I feel like this is...normal. The alternative would be that a machine tries to guess the voter's selections based off what they wrote which....seems bad.

The next section is heavily redacted. Unlike ASOG, I'm not going to try and assert something one way or the other without additional information.

Here, they point out an XML exception, and then say "Bottom line is that this is a calibration that rejects the vote (see picture below)." - that pic is redacted, but XML forbids whitespace chars in element names, so that's the likely cause of the error w3schools.com/xml/xml_elemen…

Without the picture, I can't verify if it's a "calibration that rejects the vote", but that seems like a leap to me. Dominion seems to use XML for config files, so it's likely that there's a space in an XML entity name somewhere that's causing those errors.

Giving them the benefit of the doubt, their conclusion still hinges on rogue adjudicators tossing ballots en masse or switching them to Biden, which we don't have evidence of (especially since Trump won the county)

I've addressed their next claim in a separate thread:

https://twitter.com/get_innocuous/status/1338576587780812800?s=20

The rest of the doc is redacted and not useful. So, to sum everything up - if they have conclusive data of fraud, it's not in the doc. The 68% error rate that's being touted is useless without a breakdown of WARNINGS vs ERROR, and additional context

Their fraud claims all hinge on the adjudication process. They claim that the system was intentionally misconfigured so that ballots would be rejected in bulk, and need adjudication, where they would either be thrown out or switched. But...

They failed to prove those claims (imo). The details in the doc around the number of errors are too vague to draw an actual conclusion. Their scheme also requires a human element - there would need to be adjudicators that were flipping the votes.

They claimed the adjudication logs were not present. If true, we can't verify how adjudication may have affected votes. But the lack of audit logs does not automatically mean fraud occurred.

I wouldn't be surprised if those logs were present, and ASOG failed to disclose them. Their case would be drastically weakened if the adjudication logs showed that the process was working as intended. Again, Trump won the county.

If Trump won the county, the deep state really missed the mark.

ALL OF THIS TO SAY: The report is too light on evidence to prove their claims, and the numbers that are being touted in support are, AT BEST, incorrect due to incompetence, and at worst, incorrect in an intentional effort to mislead the public.

FIN

Here's an unroll of this insanely long thread: threadreaderapp.com/thread/1338594…