M: Without trying to figure out the numbers behind it, please come up with a methodology for figuring out exactly— sorry, not exactly —figuring out about how much private and national money has been spent dealing with this election fraud bullshit. The number of people that have been drafted into it because they work for the government, the number of man hours that have been spent trying to fix a problem that doesn't exist, all of that. It's just too much for me to contemplate right now, but I can review a suggested aggregation and edit it with a lot less mental effort.
G: I’ll treat this as a cost-accounting problem, not as an argument about whether fraud exists: define the activity buckets, decide what counts, and then build one aggregation that avoids double-counting government labor, litigation, audits, security changes, and private spending.
G: [Non-text content: thoughts; content_type=thoughts]
G: [Non-text content: reasoning_recap; content_type=reasoning_recap]
G: Yes. The cleanest way to do this is to treat it almost like estimating the cost of a natural disaster: identify every stream of activity caused by the event, establish what would have happened anyway, and count only the incremental burden. That last part is crucial because the government already investigates genuine election crimes, conducts routine audits, and maintains election security systems. DOJ has a permanent Election Crimes Branch, for example, and the EAC treats ordinary audits and recount procedures as standard election-administration functions. citeturn534732search0turn534732search2
I would build the project around the question:
“What resources, public and private, were consumed in the United States because officials, organizations, or citizens had to investigate, litigate, accommodate, rebut, secure against, or otherwise respond to allegations of widespread election fraud beyond the ordinary level of election administration?”
That formulation lets you count the phenomenon without having to assume that literally every fraud allegation was false.
1. Extraordinary election administration
This is probably the most tangible category.
Count additional:
But subtract the normal baseline: legally required audits, ordinary recount procedures, routine canvassing, etc.
That distinction matters because post-election auditing is itself a normal verification mechanism; the EAC actually maintains workload tools specifically for estimating staff/time required for hand counts. citeturn534732search4turn534732search6
So:
extraordinary election cost = actual activity − activity that would normally have occurred
That becomes the basic rule for the entire project.
2. Government employee labor
I'd keep this separate from actual appropriated expenditures because it captures the thing you're particularly interested in: people being pulled away from their jobs.
For every documented episode, record:
agency × employees involved × estimated hours × fully loaded hourly compensation
Personnel could include:
DOJ's election-fraud system itself shows why you want incremental rather than total personnel: election-crime investigations already involve the Election Crimes Branch, U.S. Attorneys' Offices and investigators nationwide as part of normal government operations. citeturn534732search0turn534732search5
For personnel, report two different numbers:
Unique people touched: “At least X public employees were required to work on these matters.”
Labor consumed: “Approximately Y employee-hours/FTE-years.”
Those tell dramatically different things and shouldn't be collapsed.
3. Government legal and judicial costs
Make litigation its own bucket.
For each case:
plaintiff-government attorneys
The important methodological trick would be to count the government's defense costs even when somebody else filed the case.
A $50,000 privately financed lawsuit might have caused $200,000 of public expenditure defending election officials, producing records, conducting hearings and operating the court. Those are separate costs generated by the same incident.
4. Legislative and investigatory activity
This deserves a separate category rather than disappearing into employee salaries:
Estimate staff hours where direct budgets don't exist.
GAO's own cost-estimation guidance is useful here because it explicitly recommends defining the scope, building a work-breakdown structure, establishing assumptions, collecting data, choosing estimation methods, doing sensitivity analysis, and distinguishing estimates from actual costs. In other words, what you're contemplating can be constructed as a fairly orthodox government cost estimate rather than a rhetorical calculation. citeturn534732search3
5. Security caused specifically by election-fraud-related threats
This is potentially enormous, but I'd keep it carefully bounded.
Count documented incremental expenses for:
But do not count ordinary election cybersecurity.
CISA already treats election infrastructure as critical infrastructure and provides ongoing cybersecurity services for threats such as phishing, ransomware and DDoS attacks, so those baseline expenditures aren't properly attributable to fraud allegations. citeturn534732search1
There is, however, a much cleaner subset: CISA and FBI explicitly describe increased threats of violence against election workers following the 2020 election and say the agencies prioritized efforts in response. That gives you a defensible attribution boundary for at least some extraordinary security activity. citeturn534732search60
6. Private litigation and political spending
Then build an entirely separate private-money ledger:
Where possible, use actual expenditures from campaign-finance disclosures, court records, nonprofit filings and invoices rather than guessing attorney hours.
Crucially, I would not count donations themselves as economic cost. Count what the organizations subsequently spent. Otherwise a $10 million fundraising campaign followed by $8 million in expenditures becomes $18 million in your calculation even though $10 million was merely transferred from one owner to another.
7. Private defensive costs
This is conceptually different from political spending.
Examples:
Here again, distinguish money actually spent from employee time absorbed.
8. Opportunity cost — reported separately
This is where the true size probably gets much larger, but it is also where an estimate becomes much easier to attack.
Suppose a county clerk spends 300 hours answering extraordinary election-related records requests. Their salary was already budgeted, so perhaps the county didn't literally spend another dollar.
But 300 hours of government capacity disappeared.
I'd therefore produce:
Cash expenditure: additional dollars actually spent.
Resource cost: cash expenditure + compensated employee time diverted from other work.
Don't merge those numbers.
That lets you say something very precise later:
“Government entities incurred approximately $X in identifiable additional expenditures and devoted approximately Y million employee-hours, representing roughly $Z in labor resources.”
No claim that $Z literally left government bank accounts.
This would probably solve most of the epistemological headaches.
CORE — highly defensible
Only things directly traceable to particular fraud allegations/challenges:
extraordinary recounts + extraordinary audits + lawsuits + investigations + hearings + government labor + documented security responses.
EXPANDED — probably attributable
Add things such as unusual public-records workloads, election-worker protection, communications campaigns, additional training, employee turnover/recruitment costs, and broader administrative responses where there is strong evidence of causation.
SOCIAL COST — deliberately separate
Things like:
I wouldn't headline this third number. It is inherently much more model-dependent.
Each row represents one cost-producing event, not one organization:
| Event | Date | Jurisdiction | Trigger/allegation | Activity | Public/private | People | Hours | Direct $ | Estimated labor $ | Baseline $ | Incremental $ | Source | Confidence |
|---|
Then give every event a unique ID.
That's important because one Arizona audit might generate ten records—county employee labor, state expenditure, outside contractor, litigation, police security, records-production labor, etc.—but they're all tagged to the same event. You can examine the event later and make sure nothing was counted twice.
I'd also keep headcount, hours, and money as three separate national totals, because they're answering different questions.
So your eventual result would look structurally like:
Public money actually expended: $___
Government labor consumed: ___ million hours / ___ FTE-years
Identifiable public employees involved: at least ___
Private money actually expended: $___
Broader attributable costs: $–$
Period covered: ___
Ordinary election administration explicitly excluded.
And for every estimated quantity, use low / central / high assumptions, rather than pretending you have precision you don't. That's also consistent with GAO's recommendation that serious government cost estimates expose assumptions and perform sensitivity/risk analysis. citeturn534732search3
The one methodological choice I would be quite firm about is not starting with money. Start by constructing the inventory of things that happened because of fraud allegations. Once that inventory exists, attaching dollars and labor-hours becomes a relatively mechanical second pass. Starting from budgets would make it very easy to miss dispersed costs—or accidentally attribute ordinary election-security spending to this phenomenon. The federal government alone has put substantial funding into general election security since 2018, and much of that serves genuine cybersecurity and infrastructure purposes independent of fraud claims. citeturn534732search8
In other words, the conceptual hierarchy I'd use is event → activity → people/hours → actual expenditure → opportunity cost, with a counterfactual baseline at every stage. That should produce something you can progressively make more comprehensive without having to rethink the underlying architecture each time.
M: Okay, I can see one thing in part two that wasn't necessarily addressed. Not just the people involved, but the jobs that were left undone due to drafting in the non-expected personnel. Like you said that there was an election fraud system, right? So maybe that doesn't count, but the people who were pulled in to investigate or prosecute or otherwise deal with crap were pulled away from other tasks, ones that you at least assume are necessary if they were a government job. And when you're talking about the DOJ, that means prosecution, up to and including tax and IRS and other personnel. Because if I remember correctly, they were drafting them from everywhere. It looks like so far that you are following an otherwise pretty thorough and methodologically sound approach, where the usual occurrences are kept separate from what I'm really asking about. Because you're right. Things that already existed or are expected, even if they aren't always called for, are a lot different from what we're talking about with all of the bullshit task forces and things like that spent to investigate, or more appropriately, try and discover some sort of leprechaun-level magical way in which fraud could have occurred that hasn't appeared before. I swear, if after ten years of this level of digging someone did commit fraud to that scale for some reason without actually benefiting the people he's accusing, I feel like we need to find those people and put them in charge of other tasks, because they're really damn good at it. I mean, motherfucker has had six out of the last ten years the entirety of the government at his beck and call, spent two of the last six years, nearly two, with everyone trying to find some pinhole in which they can stake some sort of elaborate expansion to justify what they're doing, and they just can't fucking find it. I mean, it's embarrassing as well as expensive and ridiculous. I mean, the conversations we've had, just comparing the numbers of—I can't remember the official name, but the voter ballot interface errors and all of that, which vastly outnumber any discovered count of any kind of voter fraud, even the accidental kind. It's just such a fucking tragedy if I weren't having to live through the crap that is associated with it. I mean, I didn't know until today that the entirety of the state of Washington basically uses mail-in ballots, and what they're trying to do, or at least what they're claiming to do, is to try and prevent, like 0.0001% error in terms of count. And that's during a presidential year, by risking the constitutionally protected voting rights of an entire state just to begin with. In number six, I would include some kind of donation percentage. Motherfucker has been campaigning on this. This is not a minor thing that has been constant. This has been, I mean, this is right up there with the Epstein stuff, in that it was part of his campaign for four goddamn years. You can't simply discount donations. I agree with the ideal of— sorry, idea of concentric estimations. I'm not sure if the final version would be exactly what you said, but at first glance it appears about right.
G: Yes — the missing piece is displacement cost, not merely salary-value opportunity cost. A prosecutor spending 200 hours on an extraordinary election matter did not just “cost 200 hours”; some other prosecutorial, investigative, regulatory, or administrative work was delayed, redistributed, or abandoned. I’d make that a first-class category rather than burying it under labor.
G: [Non-text content: thoughts; content_type=thoughts]
G: [Non-text content: reasoning_recap; content_type=reasoning_recap]
G: Yes. I would revise the model in two important ways based on that.
First, “government labor consumed” is not enough. It needs a paired category called something like displaced government function. The distinction is:
Resource consumption: Agent/prosecutor/analyst spent 300 hours on the election-related assignment.
Functional displacement: What work was that person ordinarily employed to do, and what happened to it while those 300 hours were unavailable?
That matters especially when personnel are imported from units whose normal missions have nothing to do with elections. And there is current evidence that this is happening across organizational boundaries: DHS is reportedly increasing voter-fraud investigations through Homeland Security Investigations with assistance from USCIS investigators, including influxes of agents into multiple states. citeturn617886search6turn617886news47 IRS Criminal Investigation has also participated in at least one current voter-fraud case alongside the FBI, State Department, state police and local police; IRS-CI's ordinary mandate is primarily tax and other financial crimes. citeturn617886search0turn617886search3
So I'd modify Part 2 into this:
For every person or organizational unit pulled into an extraordinary election-related activity, record four things:
A. Who was diverted?
Agency, division, occupation and approximate number of personnel.
B. How much capacity was diverted?
Hours, workdays, or ideally FTE-months.
C. What was their normal mission?
Tax enforcement, human trafficking, counterterrorism, financial crime,
immigration adjudication, ordinary prosecution, local policing, election
administration, etc.
D. What happened to that workload?
And D needs several possible classifications rather than an assumption that every diverted hour produces an identifiable abandoned case:
That gives you something considerably better than just multiplying salary × hours.
For example, DOJ says its Criminal Tax Section exists to investigate and prosecute tax evasion and related financial crimes, often working with IRS-CI and U.S. Attorneys. citeturn617886search7 If five people normally doing that work spend substantial time on an election initiative, we should record both the election-investigation labor and the tax-enforcement capacity removed. We should not automatically claim “five tax cases went unprosecuted.” We'd look for actual changes in caseload, backlog, case-processing time, overtime or reassignment.
That also suggests an important secondary-effect layer:
Agent A is reassigned → Agent B absorbs Agent A's caseload → Agent B's work slows → supervisor spends time reallocating cases.
The first reassignment is direct displacement. The rest is cascade displacement. I'd count it when documental evidence permits, but keep it outside the most conservative core estimate.
This could produce a national result much more meaningful than a salary total:
X million federal/state/local employee-hours were diverted, equivalent to Y full-time employee-years. Of those, Z employee-years came from functions outside ordinary election administration. Documented consequences included ___ hours of overtime, ___ increase in backlog, ___ delayed matters, etc.
That is probably closer to the thing you're actually trying to measure.
The second change is your objection about donations, and I think you've identified a category I threw out too aggressively.
I still wouldn't treat every dollar donated as a dollar of social expenditure, because economically that's initially a transfer. But you're right that fundraising itself can be evidence of resources mobilized by the issue, and if election-fraud claims are materially responsible for attracting donations, omitting the money entirely loses an important dimension.
So I'd create a separate category:
Don't add this directly to “money spent.” Report it separately as:
Money raised through solicitations substantially premised upon election-fraud/election-integrity claims.
And estimate attribution using the actual fundraising appeals.
That becomes a content-analysis problem.
Take all available emails/texts/web solicitations and classify them something like:
Rather than arbitrarily deciding that a mixed email is “30% election fraud,” I'd probably initially use two bounds:
Strict fundraising total: only donations responding to solicitations whose principal appeal was election-fraud-related.
Expanded attributable total: strict total plus a modeled fraction of mixed-purpose solicitations.
Campaign committees report enormous amounts of receipts and spending to the FEC, so there is a well-defined financial universe against which this can be compared. For example, Trump's principal 2020 campaign committee reported about $744 million in receipts during 2019–2020. citeturn218477search10 The FEC separately distinguishes receipts from disbursements, which reinforces why I'd preserve “money raised because of this” and “money subsequently spent because of this” as separate quantities. citeturn218477search9
That gives you potentially three private-money measures:
Money solicited/raised using the issue
Money spent promoting or pursuing the issue
Money spent by other parties responding to the
issue
Those can overlap enormously, so they shouldn't be summed indiscriminately. But they're all legitimate measurements of its economic footprint.
And your Washington example points toward another methodological category that I think we should add: risk imposed, which should emphatically not be converted casually into dollars.
Washington really does conduct every election by mail; registered voters automatically receive ballots. citeturn218477search0turn218477search1 The state's current description of federal actions says proposed USPS requirements would impose substantial burdens on county election offices and the Secretary of State, including envelope redesign shortly before the November 2026 election. citeturn218477search2 Current litigation concerns whether federal postal requirements could prevent otherwise valid ballots from being delivered or returned; the administration describes the requirements as election-integrity measures, while challenging states and voting-rights groups argue that they threaten lawful voting. citeturn218477news37
For our accounting project, that suggests:
Count separately:
And then separately document population exposed to the procedural change—e.g., how many registered voters operate within the affected system—without turning “possible disenfranchisement” into some invented dollar figure.
That gets at something your monetary number alone can't: resources can be consumed while simultaneously introducing new failure modes into a system that was already functioning.
So at this point I'd slightly reorganize the whole framework into five top-level ledgers rather than one giant cost number:
| Ledger | What it answers |
|---|---|
| Cash cost | How many additional dollars were actually spent? |
| Labor diversion | How many employee-hours/FTE-years were consumed? |
| Displaced function | What ordinary government work was delayed, transferred or lost? |
| Private mobilization | How much money was raised and spent around the claims? |
| Institutional burden/risk | What systems had to change, defend themselves, or operate under added constraints? |
Then the core / expanded / social-cost concentric rings apply inside each ledger.
That is better than trying to force everything into one dollar figure. We might eventually be able to say, for example, “the directly identifiable cost is at least $X,” but alongside it would be the much harder-to-dismiss statement that Y full-time employee-years of governmental capacity were redirected, including Z from agencies or divisions whose normal missions were unrelated to election administration.
And I think displaced function may ultimately be one of the strongest parts of the study because it avoids the semantic argument “but those agents were going to get paid anyway.” Yes. The question is what taxpayers were already paying them to accomplish that they weren't accomplishing while reassigned.
M: A more difficult to parse version, but one which appeals to me in terms of kvetching accurately, would be just the amount of American news hours and effort that have been degraded with this. I mean, just think about the number of times that a speech was covered because it was supposed to be politically important, like just the State of the Union address. How much time and effort was not just dedicated to demonstrable bullshit, but also dragged away valuable national focus from something that actually goddamn matters.
G: [Non-text content: thoughts; content_type=thoughts]
G: [Non-text content: reasoning_recap; content_type=reasoning_recap]
G: Yes. And I think this is sufficiently different from the financial calculation that it deserves its own ledger: national attention displacement.
There is actually a measurable object underneath the kvetching. News has a finite attention budget. A nightly broadcast has only so many minutes; a newspaper homepage has only so many prominent slots; reporters and producers have finite working hours. Research on American broadcast news explicitly treats airtime this way and finds that the amount available for substantive issue coverage is limited and has declined substantially over time. citeturn0search11
So I would measure this in several layers.
1. Raw media occupation. Build a transcript database from television/cable/radio and article databases and identify every segment substantially concerned with election-fraud allegations, investigations, lawsuits, audits, voting-machine allegations, election-denial controversies, responses and fact-checking. Measure minutes of television/radio and article count/word count for print/digital.
There is evidence that this isn't remotely a hypothetical measurable quantity. One analysis found more than 21,000 separate 15-second cable-news segments mentioning election fraud in an election context between Election Day 2020 and July 2022 on CNN, Fox and MSNBC alone. Converting those clips produced about 225 aggregate days in which the subject was being discussed somewhere in those three networks. That doesn't include broadcast television, radio, newspapers, online journalism, local news, Newsmax/OAN, or anything since 2022. citeturn0search1
Even before the 2020 election, another analysis found that just 40 Trump tweets about voter fraud received 92 minutes of on-screen exposure across CNN, MSNBC and Fox during an eight-month study period. citeturn0search8
2. Journalistic production cost. Airtime dramatically understates the resource consumption. A six-minute segment might require a producer, researcher, correspondent, editor, camera crew, graphics staff, legal review, booking personnel and fact-checkers.
So every hour broadcast becomes:
one audience-hour of news capacity + X journalist-hours required to produce it.
You could estimate newsroom labor using staffing information, interviews where available, and standardized low/middle/high production assumptions. Importantly, fact-checking counts. If somebody makes an unsupported factual allegation in 30 seconds and five journalists collectively require six hours to establish and explain why the evidence doesn't support it, the resource consumption is six hours and 30 seconds, not 30 seconds.
Research examining journalistic responses to false information in the 2016 and 2020 elections specifically studied the corrective work done by major newspapers, broadcast networks, NPR/PBS and the cable networks, so there's already scholarly literature from which we could build the coding scheme. citeturn0search3
3. Opportunity cost of coverage. This is the really interesting thing you're describing, and also the hardest to estimate responsibly:
What would have occupied that journalistic capacity otherwise?
I would not try to declare retrospectively that “this 12-minute fraud segment stole seven minutes from Ukraine, three from healthcare and two from climate change.” That's unknowable.
Instead, compare news composition against a counterfactual baseline.
For instance, take comparable periods before the election-fraud issue became unusually prominent and determine the normal distribution:
international affairs / economy / healthcare / Congress / courts / science / state/local government / disasters / etc.
Then look for statistically abnormal displacement during periods when election-fraud coverage spikes.
You could consequently say something defensible like:
“During high-fraud-coverage weeks, election-fraud-related material occupied X% of available political-news airtime, while coverage of other substantive subjects collectively fell Y% relative to comparable periods.”
That is much stronger than deciding ourselves which stories “actually mattered.”
4. Event hijacking. Your State of the Union example suggests a particularly good subcategory.
Take events whose primary subject wasn't election fraud:
State of the Union
presidential press conferences
campaign policy speeches
congressional hearings
debates
interviews
governors' events
international summits
Then measure:
event airtime devoted to fraud claims + subsequent news coverage devoted to those claims + fact-checking/reaction coverage generated afterward.
That gives you an attention multiplier.
A three-minute digression in an hour-long speech might generate two hours of cumulative cable discussion, twenty newspaper stories and hundreds of journalist-hours afterward.
And there's an important distinction here between covering a claim and amplifying it. In November 2020, ABC, CBS, NBC and MSNBC actually cut away from a Trump address when he began making unsupported claims about election fraud, while CNN and Fox carried the whole address; the networks that interrupted it then explained or fact-checked what had been said. citeturn0search0 That's useful empirical evidence because it illustrates that journalists faced an actual allocation decision: continue carrying the president of the United States, interrupt him, fact-check him, or cover something else.
5. Audience-attention consumption. Then there's an absolutely gigantic number that I would calculate but never add to the financial total:
viewer-hours consumed
Ten minutes watched by 3 million Americans = 500,000 person-hours of national attention.
Do that across television audiences, streaming views, radio audiences, podcasts and perhaps article readership where reliable engagement-time estimates exist.
I'd be extremely conservative here because a television being tuned to something doesn't mean somebody devoted their undivided attention to it. But even a conservative estimate would tell us something different from newsroom expenditure:
How much aggregate American human attention did this subject occupy?
And then I think there's one final layer that captures precisely what you're annoyed about but which should remain descriptive rather than monetized:
Suppose:
That's not ten independent news events. It's one claim generating an attention cascade.
So I'd assign every descendant story the same originating-claim ID that we discussed for government expenditures.
Then we could actually reconstruct chains:
Original allegation → 37 broadcasts → 216 articles → 14 official responses → 3 investigations → investigation results → renewed claims → 81 additional broadcasts...
That makes the phenomenon you're describing empirically visible. It distinguishes news generated by an underlying occurrence from news generated because society is repeatedly processing the same assertion.
And there is research going back to 2016 showing that “rigged election” claims themselves became objects of news coverage, rather than merely underlying election-administration events being reported. citeturn0search2
So I'd now expand our overall project to include a sixth ledger:
National attention cost = journalistic labor + news-space/airtime occupation + audience-hours + displaced coverage + recursive attention cascades.
The first three can be quantified reasonably well. Displaced coverage can be statistically estimated. The last one can be mapped extraordinarily well even if we refuse to put a dollar value on it.
And that last map might actually communicate the phenomenon better than a giant dollar number: here was the original claim; here are all the government investigations, lawsuits, journalist-hours, broadcasts, articles and subsequent controversies that radiated outward from it; here is what those investigations ultimately established. That would let the reader see the opportunity cost rather than asking them simply to accept that it existed.