Live Fact Checking is Journalistic Malpractice

Bob Yentzer

September 2025

A large swath of the public, mostly Republicans, don't trust the mainstream news media. They believe it's biased. So you would think that Network and Cable News outlets would cease and desist from displays of hubris that confirm the public's worst suspicions.

Case in point: live fact-checking (LFC) during televised presidential debates.

This practice debuted in 2012 when CNN's Candy Crowley "corrected" a statement by Mitt Romney during a Town Hall with President Obama.

Unfortunately her attempt was more confusing than enlightening, and afterward she admitted as much: "[Romney's claim] was right in the main, I just think that he picked the wrong word.”

Of course, her intervention evoked a firestorm of criticism. Republicans argued that it revealed CNN's leftwing bias because only Romney was interrupted. Since Crowley failed to catch any of Obama's misstatements, he left the debate with credibility intact. So, in this sense the Republicans were right; Crowley definitely put her thumb on the scale.

The controversy chastened journalists' enthusiasm for LFC, despite their inflated self-image as arbiters of the truth. For the next 11 years debate moderators refrained from disputing the fibs of politicians. Then in March 2024, Trump reemerged as the Republican candidate for President. The orange charlatan clinched the opportunity to lie and slander his way into a second Presidential term. The prospect that the debate stage would be dominated by his torrent of delusional rhetoric was too frightening for left-leaning journalists to endure.

So, on September 10, 2024, ABC's moderators for the first Trump-Harris debate resurrected live fact-checking on the excuse that viewers needed a guardian to arbitrate the truth. But, having majored in non-STEM fields, they were unaware that the feasibility of their good intentions was rebuked by science.

Here's the rub: The science of human cognition predicts that LFC is more likely to distort viewers' perception of the candidates than sharpen it. The findings of cognitive psychology show that the execution of fact checking in real time is prone to gross random errors as well as personal bias; and because of this, it is more likely to slant the outcome of a debate than level the playing field. In other words, LFC is an inconsistent hit-or-miss game, the accuracy of which is highly unreliable.

Frequent mistakes occur because of the ways in which the brain operates on the content of semantic memory. When fact-checking involves precise details the brain switches to a more complicated mode of accessing memory, known as "strategic retrieval." This multiplies the number of cognitive operations it takes to retrieve an answer from memory. Therefore, it multiplies the opportunities for random mistakes.

The difficulty is compounded by the fact that LFC entails two distinct searches of semantic memory; first to determine whether a statement is false, and then to dredge up the correct answer. This multiplies the chances of error.

One last wrinkle: accuracy is not a binary variable. A given fact-check can be more or less accurate depending on the quality of evidence backing up a 'correction.' In the heat of the moment an ambitious moderator is likely to dispute a statement before she realizes that the evidence for a correction is lame. Rendering a worthless correction is an error.

The following analysis of the way LFC was executed during the Trump v. Harris match explains why 'journalistic malpractice' is the appropriate verdict.

Autopsy of the Trump-Harris Debate.

The limitations of human cognition were quite evident in the performance of the debate's moderators, David Muir and Lindsey Davis. Their performance clearly illustrate why attempting to screen a candidate's rhetoric in real time is inevitably plagued by error, both random and systematic. Screening invites two kinds of error: False Negatives and False Positives.

A moderator who screens for factual accuracy is like a radiologist who screens for breast cancer using mammography. If a mammogram reading detects a cancer that is really not there, the error is a false positive. But if the reading fails to detect a cancer that is actually present, it's a false negative.

People have confidence in mammogram screening because it is highly accurate: the rate of false negatives is about 13%. For false positives, the rate is 12%. By comparison, the Trump-Harris spectacle showed that live fact-checking is a juggernaut of false negatives.

False Negatives: the Bane of Accurate Fact-Checking.

Live fact checking (LFC) results in a false negative when a candidate prevaricates, but the moderator fails to challenge it.

False negatives are the biggest threat to a fair debate. LFC is most likely to bias the outcome of a debate when one candidate's falsehoods are overlooked more often that those of the opponent. The reason is obvious: Every correction issued by the moderator taints a candidate's credibility. So, a candidate whose falsehoods are regularly overlooked will appear more trustworthy than the candidate whose dishonesty is diligently corrected.

And so it was with Kamala Harris. While the two moderators called out Trump's falsehoods six times during the debate, Harris was never challenged, not even once. Six rebukes versus zero suggests a tilted playing field.

The bias toward Harris is clearer when the rate of false negatives is considered. To calculate this we need to know the total number of fallacies asserted by the candidates. This count was provided by professional fact-checkers a day or two after the debate. Systematic fact-checking by the staffs of several news organizations came up with this pooled estimate of total falsehoods: Trump=12, Harris=8.

Recall that mammogram screening is trusted because the rate of false negatives is only 13% But during the debate, moderators missed 6 of Trump's 12 falsehoods, which is a false negative rate of 50%. And they overlooked all 8 of Harris' Pinocchios, for an error rate of 100%. So, she left the debate with an aura of truthfulness she didn't deserve. What does this say about the trustworthiness of having TV stars screen for fallacies during a live debate?

False Positives

In mammogram screening, a false positive occurs when the radiologist sees a cancer that is LATER found not to exist. But screening the verbiage of debaters in real time is different, because LATER findings are irrelevant to the outcome of the debate. In the immediacy of live fact-checking, a statement is true unless the moderator (or the opponent) presents a valid correction.

Therefore, a false positive occurs when the moderator's correction of a statement fails to prove that it's false. That is, the moderator's rebuttal is factually spurious: erroneous, biased, unsubstantiated, disreputable, illogical or irrelevant. Unlike a false negative, a false positive is an error that tips the scale against the candidate because it falsely impugns his reputation.

Muir's attempts to fact check Trump provide two excellent examples.

1. In response to Trump's wild assertions about the influx of migrant criminals during Biden's "open border" policy, Muir countered with a non sequitor: "as you know, the FBI says overall violent crime is coming down in this country."

True, the FBI's pre-debate update showed that crime was declining during the first six months of 2024. But border crossings were also falling during that period, so MUIR's attempted correction actually supported the gist of Trump's argument - less immigration means fewer criminals

Trump's rant focused on the influx of "bad people" during the huge surge in immigration from 2021 through 2023. During this period, the number of violent crimes actually rose by 2 million, according to the National Crime Victimization Survey. The FBI's data on crimes reported to police showed a similar pattern.

In other words, Trump's message that Biden's reckless immigration policy increased the number of criminals in the US was a lot more truthful than Muir's rebuttal. By impugning Trump's truthfulness with an irrelevant fact, Muir committed a false positive error. Muir tipped the scale against him.

2. True to form, Trump repeated an emergent rumor that Haitian immigrants in Springfield Ohio were eating other people's. To understand why Muir's "correction" qualifies as a false positive, consider the time frame.

The pet-eating accusation first appeared in a Facebook post just 5 days before the debate. The rumor grabbed widespread attention only after JD Vance publicized it in a tweet just one day before the debate. Obviously, 24 hours was not enough time for local authorities and journalists to adequately investigate the validity of the rumor. So, by the start of the debate, there was not a shred of objective evidence that could credibly disprove the rumor.

So, Muir's attempt to correct trump was reduced to a message from the city manager that stated the obvious: "there have been no credible reports of specific claims of pets being harmed, injured or abused by individuals within the immigrant community." In other words, "evidence is absent."

Therefore, Muir's 'quotation' did not refute Trump because it violated a basic rule of logical inference: "the absence of evidence is NOT evidence of absence."

The Scorecard of LFC Errors

The scorecard shows the benefits and costs that fact checking errors inflicted on the two candidates.

False negatives benefit a candidate. They conceal a candidate's mendacity. Every time a moderator fails to correct a falsehood, it averts damage to the candidate's credibility.

False positives are a cost. Every time a moderator's falsely rebuts a statement, it smears the candidate.

A candidate's NET benefit score is simply the difference between the number of (beneficial) false negatives and (costly) false positives. As the scorecard shows, Harris benefited from fact-checking errors far more than Trump: 8 vs. 4.

Beguiled by their own inflated status, Muir and Davis rationalized that live fact checking on their part would promote a fair and balanced debate in service of the truth. Just the opposite transpired. The moderators' fact checking errors biased the outcome of the debate. Harris' 100% false negative rate fostered the impression that she was the Mother Theresa of veracity. Indeed, the moderators were infinitely (50/0) better at detecting the untruths spewed by Donald Trump than those of Kamala Harris.

This discrepancy is not an artifact of random errors but of bias. After all, the motive for resurrecting live fact checking in 2024 was to discipline Donald Trump, and by implication, defend Kamal Harris.

For example, less than 5 minutes into the debate Harris delivered a Pinocchio that was impossible to miss: "Donald Trump left us the worst unemployment [6.4%] since the Great Depression."

Yet neither one called it out, maybe because of ignorance, but more likely because they chose not to embarrass Harris.

The public trusts mammography because the error rate for false negatives is only 13%. The corresponding error rate on the debate scorecard is 70%, a sure sign that LFC is not very trustworthy. In fact, 100% percent of Harris' 8 lies went undetected (which is an accuracy rate of ZERO). The moderators also failed to rebut 50% of Trump's 12 lies, for an accuracy rate of 50%. All told, 14 out of 20 falsehoods went uncorrected, for a false negative error rate of 70%.

Final word.

Having argued that the flaws of human cognition assure that LFC tilts the playing field, I now admit that its effect in the Trump-Harris debate was nil. That's because the minority of adults who watch presidential debates are somewhat atypical: being politically engaged, they follow the news and are familiar with the issues. Most are partisans who have already chosen sides. And they tended to be much older and better educated. So, even though a large majority of the audience agreed that Harris won the debate, this had no detectable effect on Trump's favorability or support from likely voters. Most viewers were accustomed to his mendacious style of rhetoric.

But as experienced TV anchors, Muir and Davis already knew that the audience was NOT naive, and did NOT need a nabob to curate the truth. Their resurrection of LFC proved to be nothing more than an embarrassing exercise in reckless self-promotion, i.e., journalistic malpractice.

------