In the first post, we discovered a serious flaw in the recount procedure — the lack of access to a piece of absolutely critical documentation, the Ballot and Seal Certificates. Without access to these, neither the seals on the ballot boxes nor the total number of ballots can be verified, making fraud untraceable in a “recount” (making it a mere retabulation).

In a follow-up post, we found that concerns about election security are far from academic speculation. This was revealed by a statewide audit of the 2020 elections. That uncovered entire locations whose ballots were not counted, as well as the wild phenomenon of “Phantom Voters” in Irving: voters who “voted” without ever showing up (!) Unfortunately, while the “audit” by the state, which did not deserve that name, discovered this crazy factoid, it failed to get to the bottom of it. It left it documented but unexplained.

At least it was discovered, though, because the e-pollbooks printed out tapes of voters, and as a result the election workers thereby caught the “phantom voters.” Of course, for this past election cycle, Dallas County switched e-pollbook venders who provided pollbooks which do not print out voter tapes at all, so now something like “phantom voters” cannot be caught.

Not comforting!

In this post, we look at another aspect of Dallas County elections that could be greatly improved with a simple change, as we shall see and that, yet again is a critical piece of documentation that was not a part of the recount: the paper voter rosters.

The paper voter rosters are the sign-in sheet that the election clerks make of voters who come to check in to vote. Except, as we shall see, the voters do not actually sign in.

Voter rosters are in triplicate. The election judge puts one in the ballot box, one in the document bag that is turned in to Elections Department staff, and one, pink copy that goes home with the election judge, the paid volunteer who is in charge of a voting location. That pink copy must be held by the judge for 22 months.

For the critical BAPS location (where the late night vote drop happened in this race), we were able to example the pink copies.

Why did we have to track down the election judge for this though? The voter rosters were in the ballot boxes, after all. We saw them there. However, during the recount when we asked if we could look at one, the recount manager said, “No, that’s not part of the recount” and did not let us look at it.

But the paper voter rosters absolutely should be part of recounts. As we opened up the ballot boxes for each location, we were given a computer printout of the number of ballots for the District 3 race. That printout was treated as gospel truth. The counters would go through the box until they found that number of ballots … and then they would stop. But what if the computer print out was wrong, and there were more ballots in the box? The recount would never show it.

As a result, we were just retabulating votes. It was not a real recount.

The computer printout can treated as gospel truth only when access is denied to the records that would flag discrepancies.

In this case there were discrepancies. To wit: the number of voters in section 3a on the Ballot and Seal Certificate should have shown 162 voters, not 160.

What happened? Our detailed examination of the paper voter rosters for BAPS showed two voters with very similar names showed up a little after 1:00pm on election day according the paper voter roster, but only one of them was recorded as voting. The same thing happened again at a quarter to 4:00pm. In both of these cases, perhaps we were dealing with two people who were related, and one of the ballots was not recorded (which would be a problem), or perhaps it was just one person with a name that got entered twice — an error. Errors by election workers happen, which is why state law requires a retabulation only when there is discrepancy of 4.

There would surely be other discrepancies on other paper voter rosters as well. Without access to them, though, everything seems copesetic. Without a doubt, the elections department knows they are opening a can of worms if they let voter rosters be looked at in recounts. Their incentive is to get the recount over with as soon as possible with as few questions raised as possible. It’s the incentive of bureaucracies everywhere.

The recount process therefore revealed that the Dallas County Elections Department wants our elections to seem secure, without taking the obvious steps to actually make them secure. We spend tens of millions of dollars on elections in Dallas County a year … yet we do not take simple steps to actually secure our elections.

Here’s a simple change that could be made: rather than the election clerks writing down the name of the voter on the roster, (a) have the voter sign the roster, (b) put a line at the bottom for the clerk to sign and date, taking responsibility for it, and (c) make them available in recounts for inspection.

That would show fraud simply and easily. A candidate could see in a recount whenever voter signatures are in the same handwriting. As a matter of fact, that’s how Coke Stevenson’s team discovered that the ballots that put Lyndon Johnson over the top in the 1948 US Senatorial election in Texas were fraudulent. His team found the voter list where the final 200 names were names in alphabetical order, written in the same handwriting and same ink. They then went and asked those people if they had voted, and they said “no.”

We might even call them “Phantom Voters.”

Tens of millions of dollars to run our county elections and we cannot do the simplest thing that proved fraud last century.

Again, all the money and procedures seem designed to produce the illusion of secure elections, rather than taking the obvious steps necessary to actually secure them against fraud. Election workers must complete tons of steps and complicated forms to pull off an election; it’s no surprise election workers get confused. But can we not ask the voter to simply sign the paper voter roster? Also, the election clerk should have responsibility for this documentation, and sign the bottom of the form. All this would cost $0, take up zero additional time, and yet be a big boon to actual election security.

Such a change would also provide safeguards when it comes to signature verification. Currently signature verification is a joke. It is not even covered in the training of election workers. One would expect election worker training to include something along these lines: “After the voter signs, check the signature. If the signature does not match the ID, notify the judge, and here’s what you do…” But there was nothing of the sort. All signatures are to be considered valid, apparently.

Also, signatures on a screen are different from paper. People sign paper in their natural, habitual way, and there are forensic protocols to verify it. On a screen? We sign differently. A “signature” is just a ridiculous scribble.

If voters signed the paper voter roster, and that was available in the recount, signatures that looked the same should be able to be compared to their signatures on file. In this case there would be at least the threat of a more robust signature verification. This would be a deterrent for fraud.

The infuriating thing is this change would cost $0 and zero time. There is literally no reason not to do it, unless you do not want elections to be secure. We spend millions to make elections complicated and seem secure, when we could actually make them more secure with simple changes.

We need to demand those changes.

Contact: TheIrvingHerald@gmail.com (corrections/comments/news tips welcome)

© 2026 Irving Herald LLC. All rights reserved.

In the first post, we discussed how the crucial documentation for each ballot box — the Ballot and Seal Certificates — were not made available for the recount ballot boxes. That was striking, since these forms are essential for cross-checking the amount of votes in each ballot box and the seals securing them. The one we did ask for and receive, for a critical location, turned out to be grossly incomplete, with many fields missing, including total votes and the seals on the ballot box. Without this documentation, there would be no way to trace if the ballot box had been tampered with, with ballots added or subtracted.

Is this a far fetched notion though? Perhaps thinking this way is the result of an overheated imagination in the wake of learning certain things, like the fact that the McDonald’s Monopoly game of our youth was completely rigged, or that Jeffrey Epstein won the Oklahoma Lottery? Surely such things don’t happen here in Dallas, right?

Yet consider the audit done by the Texas Secretary of State of the 2020 General Election in Collin, Dallas, Harris, and Tarrant counties. It makes for interesting reading.

This audit showed a lot of wild things, including some things close to home. For instance, on Election Day 2020, at Thomas Haley Elementary School in Irving, there were 499 ballots cast but only 354 check ins (page 82). What the heck?

At least this was not as bad as an Oak Cliff polling location, where 233 people voted with only 156 check ins … but, even more concerning, none of those votes even got recorded. Why weren’t the ballots counted? The report leaves it an unexplained mystery (page 98-103).

So, votes can apparently just disappear, and even a state audit is unable to get to the bottom of it. Deleting a few voters here and there to get to a desired outcome could potentially pass unnoticed.

Unless you looked at the audit logs for the machines. The voting machines, tabulation machines, and pollbooks all generate detailed audit logs of all actions and events that have occurred on the unit, which show every action and event, including access attempts, access of system functions and errors, with their timestamps. The audit logs were not a part of our recount; but shouldn’t they be? When a candidate is spending thousands of dollars on a recount, it seems like there should be automatic access to things like the paper voter rosters, Ballot and Seal Certificates, and audit logs. Why wouldn’t there be?

In the state audit of the 2020 election in Dallas, though, the investigators did not look at the audit logs either, but just focused on election records and procedures, identifying discrepancies without reconstructing the system events with a forensic review.

The failure of the state audit was disappointing, but perhaps it was too big a task to do a full forensic audit of the four largest counties in Texas. The Secretary of State’s Office had their hands full just getting Harris County to fork over their records.

Not that Dallas did such a great job locating and turning over records either. The audit report has this to say about early voting at the Irving Arts Center:

The data for Irving Arts Center regarding check-ins contained significant discrepancies, largely due to the absence of records. For example, while the Daily Report Form and Handwritten Daily Voter Roster reflected approximately 11,500 voters checked in during Early Voting, the available pollbook tape records only showed 11,119 voters. FAD [the Forensic Audit Division of the Texas Secretary of State] is unable to account for the additional voters was due to tapes missing from the records. (Page 94)

Now for something really wild…

From the report:

In reviewing the records associated with Irving Arts Center, FAD observed a log that appeared on October 16, 2020 entitled the “Phantom Voter List.”

The list was tracking a strange phenomenon: voters were voting who did not show up to the polls.

The log reflects that there were names printed on the electronic pollbook tape that did not appear on the handwritten daily roster. An inspection of the tapes and handwritten rosters confirmed this was occurring. (Page 95)

Crazy! It was not just a glitch here or there, but so pervasive that someone created a form to track it. It was entitled: “Phantom Voter List.”

What in the world?? Who even created this form? Nobody seems to know:

Dallas County located some emails between members of their staff and ES&S [Election Systems & Software, which runs the pollbooks, voting machines, and tabulators] reporting the problem. Dallas County believed the problem had been fixed at some point but could not remember when or how. Dallas County also believed this form was used to keep track of what was occurring so that the proper voters were getting recorded as having vote history, instead of the “phantom voters” being printed on the tapes. FAD interviewed ES&S about the issue. ES&S understood the issue to be related to a non-unique identifier having been used when the list of registered voters was uploaded to the pollbook. ES&S also expressed they believed the problem had been resolved at some point, but was unsure how or when. FAD also interviewed VOTEC, Dallas County’s VEMACS voter registration system vendor. VOTEC explained that in preparing for an election, Dallas County uses the information from its voter registration system to export a list of registered voters for upload to the electronic pollbook. Periodically, the county will query the voter registration system to see if any changes or updates have been made that need to be made for the pollbook. VOTEC explained that after the export occurs, they have no insight into how the data is imported into the electronic pollbook. Voter records that are exported only contain two numerical unique identifiers: the driver’s license number and the voter’s VUID. VOTEC said they had heard of this occurring in other circumstances but were unsure of how or why. VOTEC said they reviewed their communications with Dallas County and there was no record of this issue that they were able to locate. (Pages 96-97, emphasis added.)

Goodness.

Nobody really knows anything apparently, and the Forensic Audit Division’s report just documented the problem without any attempt to get to the bottom of it:

The electronic pollbooks were producing unreliable records and for some days the handwritten rosters were missing entirely. It was not possible to fully reconcile the check-ins versus ballots cast at this location due to missing, incomplete, inconsistent, or unreliable records. (Page 97)

At least the issue got flagged so we know about it. But of course the reason it got flagged was because the pollbooks printed out tapes. Dallas County was praised in the report for this:

Dallas County’s practice of printing daily pollbook tapes – showing the names of voters – combined with their handwritten check-in log made the “phantom voter” issue easier to track. (Page 12)

But now, after switching pollbook vendors for this year’s election, we no longer have daily printed pollbook tapes. So what was used to flag “phantom voters” no longer exists. Problem solved!

In the next post, we will talk about another critical piece of documentation that was not available for us to review during the recount, but that we managed to track down, examine, and which revealed further discrepancies: the paper voter rosters.

Stay tuned.

Contact: TheIrvingHerald@gmail.com (corrections/comments/news tips welcome)

© 2026 Irving Herald LLC. All rights reserved.

In these detailed follow-up posts we’ll get in the weeds, in order to share some takeaways and analysis, and hopefully inform any candidates in the future considering or undertaking a recount.

This race was Abdul Khabeer, Kejal Patel, and Tammam Alwan, none of whom got an outright majority in the general election. We covered this thrilling race here.

The race went to a runoff between Khabeer and Patel, however ultimately only five votes separated Patel and Alwan, the latter calling for a recount.

The recount occurred over two days — Tuesday, May 12 and Friday May 15, at Central Count — at the Dallas County Elections Department at 1520 Round Table Dr., in Dallas.

The procedures to request a recount are here, the Secretary of State’s Recount Guide is here, and the law governing recounts is here.

The first day was Early Voting and Mail In Ballots; the second day involved counting Election Day ballots and reconciling them.

Each ballot box is opened and the ballots sorted to find the particular race of interest. There is a computer printout — reports from the Central Accumulator, the main tabulation computer — that states the number of ballots cast for that race at that location. Once the counters get to that number, they stop, and that is why this process should not be called a “recount” but rather a “retabulation.”

To see what the process looked like, here’s video of the livestream:

An erstwhile District 6 City Council candidate has a cameo
serving as a watcher for the Kejal Patel campaign

You’ll see the watchers observing the counters. For candidates doing future recounts, the best practice for watchers is to record results with tally marks as they are called out in a table with candidates forming the columns and ballot styles the rows.

The video stream has no sound, does not really show what is happening, and is not archived online. Cell phones are not allowed in the room. It is unclear why that should be the rule except to make it difficult to document what occurred.

Importantly, we were not given the Ballot and Seal Certificates for each ballot box. We asked for one and it was provided, however, all of them should have been available and ready as each ballot box was examined. That way, we could cross-check the seal numbers on the boxes as they were being broken into with what the seal numbers said on the form. When asked, the Deputy Elections Administrator Malissa Kouba said they weren’t done being processed yet; she did not explain exactly what processing still hadn’t happened ten days after the election.

Here is the one we asked for and received, for the BAPS voting location:

You can see there are problems with the way the judge filled it out. Sections are blank that should be “0” or “N/A” (sections 2a, 2c, 3b, 3c, 3e). More importantly: (a) the signature of the presiding judge omitted the date and time, (b) the clerk did not sign it at all, (c) section 3a, “Number of voters recorded on the Voter Roster” was left blank, and (d) section 2b, “Public Count number from the tabulator screen” was also blank.

Those last two are key numbers, showing the number of ballots counted by the tabulator at closing, and the number of voters as recorded in the paper voter roster.

Why was this form accepted by Elections Department staff when it was dropped off?

Not only did they accept it, they failed to record the seal number for the ballot box. So in our recount, even after requesting and getting this form, it was not possible to verify that the seal on the ballot box was indeed the seal that was put on the box by the election judge. Likewise, there was no seal number for the pouch that contains the USB drive.

This is unnerving. In this circumstance, anyone with an admin password and access to an ePollbook and tabulator at central count could vote as many ballots as they wanted, add them to the ballot box, and this fraud would be untraceable in any recount.

Also, if the vote totals are not recorded, there’s no way to follow state law that mandates retabulating ballots in case of discrepancies.

As we heard on our podcast on this topic, the response the candidate requesting the recount gave when he was told of this law was, “How is it OK to have a three vote discrepancy, when making the runoff was decided by only five votes!?”

He has a point. There were 34 voting locations on Election Day. If each of them is off by 3, that is a total of 112 votes, and in low-turnout municipal district election, that can be a massive number. Anyone with access to the system could delete the votes of three voters for a candidate in each location, controlling any tight race in a way unable to be traced, since a recount — really just a retabulation of the computer’s result, in the absence of access to see the Ballot and Seal Certificates, the paper Voter Rosters, and the audit logs — would not be capable of revealing any discrepancy. Recall that the “recount” just relies upon what the computer says; as soon as the counters get the result the computer printout says is true, they stop. If there is no ability to cross-check the computer numbers, there would be no way to detect this type of fraud.

Contact: TheIrvingHerald@gmail.com (corrections/comments/news tips welcome)

© 2026 Irving Herald LLC. All rights reserved.

Would you vote for that candidate?

No. Rather you would think, "What’s going on here?”

Artificial Intelligence is a brand new technology but Americans are already sick of the cheating and the fakery.

A mayoral candidate in Wyoming suggested turning local government over to an AI chatbot … and got only 2% of the vote. Politicians are using ChatGPT to write their speeches, with absurd results (including British MPs all sounding like Americans).

The idea of an “AI Candidate” does not yet have a face to go with it in popular consciousness.

Vivek Ramaswamy was mocked online as the AI candidate in 2024; however, it was because his answers were too scripted and fluid.

A candidate who was the opposite, who literally has nothing to say apart from ChatGPT would be a new one. To actually elect a candidate like this would be to invite mockery, and to deserve it.

Well, guess what? Irving is on the verge of doing just that:

This was the very first question in the April candidate forum. The question was, “introduce yourself and tell why you’re running.” Irving City Council District 3 candidate Kejal Patel could not answer it.

She is a deer in headlights. She does not seem to be having a good time campaigning. She would likely not enjoy serving on our city council.

And how could she even serve, without the language skills to understand what is going on, or the ability to effectively express herself?

The times when she is not reading answers, she says things like

“My story is rotating hard work and deeply commitment”

or

“We need safety neighborhoods with the childrens and the seniors always.”

However, when looking down and reading, Patel’s answers are in flawless English.

The safe bet is that the answers she’s giving in these cases are written by AI.

No mention here that her daughter goes to Coppell ISD,
even though she pays Irving ISD property taxes

Was she being fed AI answers at the candidate forum?

Yes, she was. After a question would be asked, a person in the audience would prompt an AI tool and copy-and-paste the answer into a note shared with her phone.

She would then read it into the mic.

Take the answer she gave above, for instance:

Don’t believe the AI checkers with their false positives; read her answer yourself. Is this her voice, or is it AI:

Whether or not it’s AI-generated, it certainly differs from the one sentence of Patel’s when she broke off reading, looked up, and spoke straight to the audience:

“I think the stronger schools bring the most stronger communities.”

When Patel is looking up, she talks very differently from when she is reading.

For example, when asked for her biggest priority, she is looking up:

However, when she just looks down and reads, she sounds like AI.

This is even to the point where, in a direct question about conservation, she pronounced that word in her answer as “conversation”:

One more example:

Kejal Patel was the only city council candidate to not do an interview with us. It is obvious why she didn’t. She has no answers of her own.

After the Dallas Morning News met with her, they wrote a whole article praising the incumbent and the other challenger in the race, but had only one sentence to say about her at the end:

Obviously, this whole matter is painful.

Unqualified candidates run for office all the time, though, and even get elected. That’s our republican system of government. We survive.

However such events are normally upsets and should be wacky upsets. This is different. This is the rallying of the whole Irving establishment behind the candidacy of someone who could turn us into a laughingstock as electing the first ChatGPT public official in the country. This is not funny.

We are the 12th biggest city in Texas and a top-100 city in the United States of America. We deserve better than what our establishment is giving us.

At least the Dallas establishment did not back Amber Givens.

What is going on here?

Contact: TheIrvingHerald@gmail.com (corrections/comments/news tips welcome)

© 2026 Irving Herald LLC. All rights reserved.

Subscribe now

It was an exciting night of watching returns last night. Congrats to all the candidates — y’all ran great races. We need good people to run.

Mayor

Not because of the mayor’s race though, which Al Zapanta won as expected. Zhanae Jackson was a surprise vote getter, netting 18% citywide and coming in second place.

In last year’s citywide race 11,054 ballots were cast, so this is a drop of 19%.

Congrats to our mayor elect, Al Zapanta!

District 5

This one was also not surprising, as councilman Mark Cronenwett was unopposed.

District 6 Special Election

The special election in District 6 to fill the last year of Al Zapanta’s term was tight, with Dana Collins giving Fahad Ahmed a run for his money.

If Ahmed had lost just 18 votes, this would have gone to a runoff.

We might see a rematch of this race in a year.

District 3

This one was thrilling. Tammam Alwan had the lead after early voting:

But then when election day results came in, incumbent councilman Abdul Khabeer passed him to take first place

As the night went on, it Kejal Patel was fading and it looked like she was not going to advance to the runoff:

…until the midnight drop, when she gained 104 votes and Alwan gained only 8:

Check out this Joe Biden-level late-night comeback:

No word year on whether Alwan will be requesting a recount. The deadline for that is 2:00pm Tuesday.

As the second-place finisher, Patel will now get the chance to unseat the incumbent Khabeer in the runoff election.

The runoff is Saturday, June 13 — early voting will run Monday, June 1 - Tuesday, June 9.

Subscribe now

Irving ISD School Board District 1

This district covers the area around University Park and Irving Heights:

Irving ISD School Board District 2

This district covers the area around Northgate and north of N. O’Connor:

DART Withdrawal Elections

Of the three referendums to leave DART only Highland Park voted to leave:

Subscribe now

Contact: TheIrvingHerald@gmail.com (corrections/comments/news tips welcome)

© 2026 Irving Herald LLC. All rights reserved.

We have recommendations here and below are all the interviews we did with the candidates.

Mayor’s Race

Al Zapanta

Zhanae Jackson

Olivia Novelo Abreu

Subscribe now

District 3

Tammam Alwan

Abdul Khabeer

Kejal Patel

The only candidate who declined to be interviewed. This video about her candidacy is going viral.

Subscribe now

District 6 (Special Election)

Dana Collins

Fahad Ahmed

Subscribe now

Our overview of the election:

Note on this episode: District 6 candidate Dana Collins messaged us and clarified his views on an important issue, so we issued this clarification here.

Contact: TheIrvingHerald@gmail.com (corrections/comments/news tips welcome)

© 2026 Irving Herald LLC. All rights reserved.

Subscribe now

Contact: TheIrvingHerald@gmail.com (corrections/comments/news tips welcome)
© 2026 Irving Herald LLC. All rights reserved.

Subscribe now

Contact: TheIrvingHerald@gmail.com (corrections/comments/news tips welcome)
© 2026 Irving Herald LLC. All rights reserved.

Subscribe now

Contact: TheIrvingHerald@gmail.com (corrections/comments/news tips welcome)
© 2026 Irving Herald LLC. All rights reserved.

Subscribe now

Contact: TheIrvingHerald@gmail.com (corrections/comments/news tips welcome)
© 2026 Irving Herald LLC. All rights reserved.

Subscribe now

Contact: TheIrvingHerald@gmail.com (corrections/comments/news tips welcome)
© 2026 Irving Herald LLC. All rights reserved.

Contact: TheIrvingHerald@gmail.com (corrections/comments/news tips welcome)
© 2026 Irving Herald LLC. All rights reserved.

Contact: TheIrvingHerald@gmail.com (corrections/comments/news tips welcome)

© 2026 Irving Herald LLC. All rights reserved.

Contact: TheIrvingHerald@gmail.com (corrections/comments/news tips welcome)

© 2026 Irving Herald LLC. All rights reserved.

This week’s episode brought to you by the Waterway Cafe:

Of note in the episode:

• Dallas has control of the DART board, with 8 votes. It spends $420 million on DART and gets nearly $700 million in services.

• Grapevine and Coppell are served by DART’s new Silver Line train, but they are not member cities. They get a la carte service. (Why can’t Irving get that?)

• The Silver Line was originally pitched as costing around $1.3 billion, but costs have ballooned to $2 billion.

• Taxes in Irving work out to $400 per citizen for DART and $75 per citizen for road repair.

• Users of GoLink in Irving have to transfer if they go to a different zone. (Why can’t Irving be all one zone?)

• Irving City Council voted 9-0 to hold an election to pull out of DART. This can only happen every six years. If it fails or the election is pulled, the next window for a withdrawal election is 2032.

At the end of the Dec. 11 work session, Mayor Stopfer listed demands we have of DART:

1. Equal promotion for Irving events in regional marketing

2. DART participating directly in our TIF District

3. Governance reform

4. DART to design and fund two remaining stations in Las Colinas (Texas Stadium and at Teleport Boulevard near SH 114)

5. Money for a Circulator bus in the Urban Center to get people to the station

Bonus quote. At the end of the Dec. 11 work session, Mayor Stopfer asked the following:

“Who is DART and what is DART? Who owns it? Are we partners? Are the 13 cities partners in DART? Which means if it’s worth… $10 billion… then we should have some equity. If we’re just a cash cow who’s just paying and we’re not gaining anything in return, then we should have the same opportunity… to pick and choose our services and determine if we want it from DART or if we want it from somebody else.”

Subscribe now

Contact: TheIrvingHerald@gmail.com (corrections/comments/tips welcome)
© 2025 Irving Herald LLC. All rights reserved.