Jake Scott, M.D. Fails the Mirror Test

His public assertions and papers make unfounded assumptions, assertions and conclusions that incorrect in ways that can harm patients. We break his record down.

Jake Scott is touted by The Guardian as an “infectious disease specialist” who is “nationally recognized” as a vaccine expert.

Scott has seven papers that mention vaccines. Exactly zero present new data on vaccination safety or efficacy. Jake has published two editorials (opinion pieces) and is a co-author on a systematic review. I do not know what it takes to become an expert in vaccines, but so far, Jake seems to be inheriting the mantel from Offit. And he is much, much worse at being smooth about it than Offit.

The Mirror Test: When Doctors Who Vaccinate Fail Their Own Standards

In the article designed to attempt to discredit ACIP member, Retsef Levi, Dr. Jake Scott tries to sow fear and warn that the new ACIP will launder predetermined conclusions through institutional authority. That claim, made by Scott and published by The Guardian, deserves the same evidential scrutiny he demands of others — because the double standard runs deeper than he acknowledges, and in more than one direction.

If both Scott and The Guardian fail the mirror test here, they should expect expect to lose credibility in the eyes of professionals and the public alike. Because the only thing worse than a hypocrite is a lying hypocrite.

Let’s remember that the ACIP of old refused to mention the word risk, and even held a vote first before any discussion of risk, failed to act on evidence of myocarditis and clotting disorders before COVID-19 mRNA vaccine recommendations, and acted as a rubber-stamp committee heedless to public comments about mounting evidence of concern over the reliance on post-market surveillance retrospective studies for vaccine safety.

Scott’s Peer-Reviewed Papers, Public Statements, and Contemporaneous Evidence
March 2026

In March 2026, Stanford infectious disease specialist Dr. Jake Scott appeared in The Guardian warning the public about Retsef Levi, the MIT operations professor appointed by Robert F. Kennedy Jr. to ACIP’s special Covid-19 immunizations work group. Scott characterized Levi’s work as using “study designs that can generate associations but cannot establish causation” and having findings “amplified as evidence” beyond what the underlying methods support. He does not inform the public that “public health” created exactly that logic trap with passive vaccine adverse events reporting systems. And he fails to represent observational research on such data sources, when conducted in earnest, as an attempt to detect signals of risk that should then be followed up on. He’s just fine with a hobbled vaccine risk system. Why? because then he can say “people are publishing results based on study designs that can generate associations but cannot establish causation”. Let’s call that “enforcing the stalement”.

He also implied that the ACIP might use unreviewed claims to justify restricting vaccine recommendations. And he described the pre-Kennedy ACIP — the one Kennedy fired and replaced — as the “international gold standard for vaccine decision-making.”

Before examining Scott’s record, the source of this article’s framing requires explicit acknowledgment. The Guardian piece is not a neutral account. It is an adversarial profile of Levi, also known as a “hit piece”, constructed around testimony from scientists who oppose his appointment and quotes from Scott, who opposes it. That framing shapes what claims get examined, which critics get quoted, and how evidence is characterized.

A prior version of this analysis accepted too much of that framing uncritically — most tellingly in its treatment of Levi’s post-publication correction, which the Guardian presents as damning and which is, in fact, the scientific process at work. A correction issued by Levi in response to criticism, where the subsequent retraction request was reviewed and denied by the journal, is not evidence of a researcher’s unreliability. It is evidence of a researcher engaging with criticism via rational discourse. Treating it otherwise requires an ad hoc, capricious and arbitrary standard that, applied consistently as triple jeopardy, would implicate most active researchers.

The Mirror Test

What follows examines Scott’s own record by the standards he invokes — and examines his institutional claims with the same skepticism he applies to his opponents’ institutional claims. The subject is not whether Scott’s position on vaccine safety and effectiveness is mainstream. The subject is whether his application of evidential and procedural standards is consistent — or whether it changes depending on which side of the debate the target is on.

Part I · The Public Claim

What Scott Said, and What It Would Require

Speaking to The Guardian in March 2026, Scott made the following statement about his target:

Claim Under Scrutiny
“What concerns me is not that someone is reviewing the data, but that the person who is reviewing it has said mRNA vaccines should be removed from the market. The question becomes, is this a genuine scientific inquiry or a review with a pre-determined decision?”

This may appear to be a reasonable standard for a scientific reviewer. A person who has publicly committed to a conclusion before reviewing the evidence cannot be treated as neutral. The standard is clear, testable, and important.

However, it must also applies symmetrically. Scott himself has taken repeated public positions in favor of broad vaccination, easier Covid-vaccine access, and preservation of the childhood schedule. Those positions are not secret. They are on record in peer-reviewed papers, op-eds, and public statements. By his own criterion, Scott is not a neutral reviewer either. He has a pre-determined position. There is no epistemic difference. We can take it one further: According to this logic, he should not be commenting if biased peoples’ positions will bias what they say.

The Guardian article fails to note this symmetry, forgetting that rational discourse requires discourse. It presents Scott as a neutral umpire calling balls and strikes. That presentation is a distortion by any measure. Scott is a player in the game he is umpiring. He has a visible prior. That prior may be represented by the main stream as better supported by the evidence than Levi’s, but appearances can be deceiving. We know about the intentional biasing of vaccine safety studies to prevent vaccine skepticism. Scott’s prior is a prior all the same. And by Scott’s own logic, it creates the appearance of a “pre-determined decision” — the thing he says disqualifies Levi.

Evidentiary Standard Slippage
Scott’s prior is not hidden. But he casual use of anecdote in favor of his own position is another offense of the double standard. In a June 2025 STAT op-ed, he wrote: “Nearly every critically ill patient I take care of in the hospital is unvaccinated.” In the Guardian, he said: “Nearly all of the patients I lost to Covid in 2021 were unvaccinated.” These are strong, memorable claims. They are also undocumented. They are not from a peer-reviewed study with a defined cohort, denominators, verified vaccination status, age and comorbidity adjustment, or independent review. They are retrospective clinical memory, offered as if they have policy weight. Scott’s own comments fit exactly his description of “strong public claims without adequate evidence”. Scott fails on his own standard, and The Guardian allowed that obvious wart to stay on the face of their article.

CDC surveillance from 2021 tells a different story. From April 4 to July 17, 2021 (13 jurisdictions), there were 616 vaccinated Covid-associated deaths. From April 4 to December 4, 2021 (25 jurisdictions), there were 22,567.

To be fair, the source (Johnson et al. MMWR, January 2022), drew on 25 U.S. jurisdictions across April–December 2021, reported that age-standardized weekly death rates were consistently higher among unvaccinated adults than vaccinated adults across all periods — with unvaccinated rates ranging from 1.5 to 11.4 per 100,000 and vaccinated rates ranging from 0.1 to 0.7 per 100,000. Those are ranges across the full surveillance window, not averages for any single sub-period. The 11.4 figure represents the peak — August, at the height of the Delta wave — not the average for July through November.

The period-specific figures for October–November 2021 are more precise: the unvaccinated death rate was 7.8 per 100,000 versus 0.6 per 100,000 among unboostered vaccinated adults, a mortality rate ratio of 12.7. Against booster recipients the ratio was 53.2. Vaccine effectiveness against death during the full Delta predominance period was approximately 94%.

The protection differential, however, is not based on experiment, which allows the establishment of causality. It is based on observational data, the same type Scott criticized Levi of using. It’s also based on case definitions that have been determined to be dependent on a fatally flawed use of RT-PCR due to false positives (recall the use of arbitrary cycle thresholds). Further, reporting of biasing causes of death to COVID-19 for the unvaccinated and to anything other than COVID-19 for the vaccinated were everywhere during that period. The CDC even announced that they only wanted residual material from unvaccinated patients for new variant determination, only to cover that up with the dead-end PDF replacement. Institutional bias analyzed uncritically by MMWR is no way to base relative rates of mortality.

The fact is that based on COVID-era public health’s best estimate, the absolute death count establishes the other necessary bound: 22,567 vaccinated adults died of COVID-19 in these 25 jurisdictions between April 4 and December 4, 2021. That number is not a negligible asterisk. It means that the category of “vaccinated person who died of COVID” was populated throughout the year, including during the Delta peak, including among the elderly and medically fragile populations most likely to reach an infectious disease specialist. The rate differential was large. The absolute number of vaccinated deaths was not zero. Scott’s clinical narrative acknowledges neither the denominator nor the case-mix that would make “nearly all of my patients who died were unvaccinated” an auditable rather than a rhetorical claim.

One element of the statement survives neither the rate data nor the “my practice” qualifier: “who would not have died had they been vaccinated.” This is a causal counterfactual applied to specific individuals NOT BASED ON SCIENCE. But The Guardian happily allows that comment to go unchallened. A 94% vaccine effectiveness against death at the population level means that vaccinated individuals still died. It does not mean that any specific unvaccinated patient or any specific collection of patients would have survived with vaccination. The claim conflates population-level statistical protection with individual- and practice-level certainty. That conflation is precisely the analytical error Scott attributes to Levi throughout the same article.

Vaccinated deaths did occur. They were not zero. They were not “nearly zero.” They were rare relative to unvaccinated deaths — but rare is not the claim Scott made.

CDC 2021 Surveillance Data (Johnson et al., MMWR 71:4, Jan. 28, 2022 — 25 U.S. jurisdictions)

Scott’s narrative may describe part of his own practice. It may even be accurate for his caseload. But it is not public-health evidence. It is an anecdote. And anecdotes, by Scott’s own logic, cannot be amplified as policy arguments without the evidentiary apparatus that would make them reliable. The Guardian lets him do exactly that — right after he criticizes the other side for using “claims that are not peer-reviewed or publicly released.”

The Precise Formulation

What an Evidence-Calibrated Version of Scott’s Statement Would Say
Jake Scott is not a neutral reviewer by his own standard. He has taken repeated public positions in favor of broad vaccination and against limiting Covid-vaccine access, while also warning that vaccine policy should not be driven by predetermined conclusions or by claims that are unpublished and unreviewed.

Yet he then offers his own unevaluated personal caseload story — that nearly all of the Covid patients he lost in 2021 were unvaccinated — as if it has policy value. CDC surveillance shows vaccinated Covid deaths did occur in 2021, even though death rates were approximately 16 times higher among the unvaccinated. His anecdote may describe part of his own practice. It cannot be treated as public-health evidence without a defined cohort, denominator, verified vaccination status, age and comorbidity adjustment, and independent review.

“What concerns me is not that someone is reviewing the data, but that the person who is reviewing it has said mRNA vaccines should be removed from the market. The question becomes, is this a genuine scientific inquiry or a review with a pre-determined decision?”

The sharpest double standard in the Guardian article is structural. Scott warns that ACIP may use “claims that are not peer-reviewed or publicly released” to justify restrictions on vaccine recommendations. The article then immediately gives him room to advance his own personal caseload narrative as if it carries equivalent policy weight. It does not, by his own stated standard. A public-health claim requires a defined cohort, denominators, verified vaccination status, age and comorbidity adjustment, and independent review. Scott’s own retrospective clinical memory has none of these properties.

Part II · Jake Scott’s Published Record

Paper-By-Paper Reasoning Failures

Scott’s published record contains multiple papers with material failures of logic, reasoning, and intellectual forthrightness. The analysis below ranks them in descending order of combined concern score, using the framework from the underlying epistemological critique. Scores evaluate logical integrity (conclusions warranted by data), engagement with contrary evidence, and rhetorical proportionality (language calibrated to evidential strength). The focus is on process failures — not on whether the paper was “ultimately wrong,” but on whether the reasoning itself was compromised at the time of publication.

He has a penchant for relying on selective attention and stating opinion as final, rigorous conclusions even when spirited debate rages.

#1. Covid-19 vaccination: evidence of waning immunity is overstated (BMJ, September 23, 2021)
Combined Concern Score: 13 (Logical Integrity: 4; Engagement with Contrary Evidence: 4; Rhetorical Proportionality: 5)
Summary of Reasoning Failure: The editorial selectively dismisses emerging evidence of waning immunity by attributing it primarily to confounders and behavioral factors, while overgeneralizing “sustained” protection against severe disease from limited studies; it fails to substantively engage with contemporaneous preprints showing clear waning signals (e.g., Goldberg et al., August 2021 medRxiv, reporting increased infection rates over time post-vaccination) and constructs a false dichotomy between boosters and global equity, ignoring hybrid policy approaches available at the time. Additional overreaches include a speculative causal claim linking waning discussions to reduced vaccine confidence without evidence, an unsubstantiated comparison deeming variants a “greater threat” than waning without metrics, a dismissive interpretation of booster immunology overstating certainty from limited data, selective reinterpretation of U.S. studies (e.g., San Diego infections attributed solely to prevalence without analysis), an authoritative dismissal of booster advocacy as “misguided” without engaging pro-booster arguments, and a logical leap demanding undefined “robust” data for boosters while shifting goalposts from accepted antibody correlates.

Most Egregious Passage
“Studies with systematic sampling do seem to suggest a modest waning in protection against infection over time. However, the primary objective of covid-19 vaccines is to protect against severe illness rather than infection, and multiple well designed studies have found sustained vaccine effectiveness against severe covid-19 for most adults.”

Forthright Version
“While some studies suggest modest waning in protection against infection, others, including recent preprints, indicate more substantial declines; vaccine effectiveness against severe disease appears sustained in available data, but long-term monitoring is needed given emerging signals of reduction over time.”

Harm Vector: Public health officials and clinicians in high-income countries might delay booster recommendations, leading to lower uptake among at-risk adults (e.g., elderly or immunocompromised), increasing breakthrough infections, hospitalizations (estimated 10-20% higher risk without boosters per contemporaneous models), and mortality during Delta waves.

Characterization: Selective attention—authors cite contrary evidence (e.g., Goldberg preprint) but downplay it through unquantified “confounding” claims without rigorous rebuttal, aligning with a pro-equity stance that prioritizes primary vaccination over addressing waning.

This editorial’s declarative title and verdict-style language foreclosed debate at a critical juncture, potentially contributing to hesitancy that prolonged surges.

#2. ESR and CRP: “It’s Time to Stop the Zombie Tests” (Clinical Microbiology and Infection, 2025)

Combined Concern Score: 13
Logical Integrity: 4 · Engagement with Contrary Evidence: 4 · Rhetorical Proportionality: 5

Scott’s invited commentary arguing that erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP) should be abandoned from clinical practice illustrates the same pattern seen in the BMJ editorial: a memorable rhetorical device deployed in place of careful evidential reasoning.

The piece introduces the phrase “zombie tests” to describe ESR and CRP, arguing that these laboratory markers persist in clinical use despite lacking diagnostic value. The underlying observation — that both tests have low diagnostic specificity — is not controversial. The reasoning error occurs in the step that follows. The commentary treats low specificity as if it were equivalent to zero clinical utility.

That leap does not hold.

Many laboratory markers are not primarily diagnostic tools but monitoring tools. CRP in particular has a documented role in guiding antimicrobial stewardship, especially in determining when antibiotic therapy can be safely discontinued. Multiple systematic reviews available before the commentary’s publication reported that CRP-guided discontinuation protocols shorten antibiotic courses without increasing recurrence or mortality.

The commentary does not engage that literature.

More strikingly, the argument conflicts with Scott’s own earlier work. In a paper published in the same journal two years earlier, Scott and co-authors explicitly endorsed CRP as a useful adjunct in antimicrobial therapy management. The 2025 commentary neither acknowledges this position nor explains the change in reasoning. The reader is left with a categorical conclusion that is not reconciled with the author’s own prior analysis.

The rhetorical centerpiece of the article is the line:

“ESR and CRP are zombie tests: they refuse to die despite a lack of brains.”

The phrase is effective as rhetoric. It is not effective as scientific reasoning. A more forthright formulation would recognize that ESR has limited clinical value in many contexts and is often overused, while CRP retains documented utility in specific applications, particularly antimicrobial stewardship and monitoring inflammatory disease activity. The evidence supports selective use and reduction of indiscriminate ordering. It does not support elimination.

The problem here is not that the commentary raises concerns about test overuse. That concern is legitimate and widely shared. The problem is the replacement of nuance with a categorical conclusion. The metaphor defeats the evidence.

If clinicians were to adopt the argument literally, the consequences would not be theoretical. CRP-guided antibiotic stewardship is one of the few inexpensive tools available to reduce unnecessary antibiotic exposure in hospitalized patients. Removing that signal could encourage longer antibiotic courses, increasing risks of drug toxicity, Clostridioides difficile infection, and antimicrobial resistance pressure across healthcare systems.

The deeper issue is the same asymmetry seen elsewhere in Scott’s record. Evidence that complicates the argument is not engaged. Prior work pointing in the opposite direction is not acknowledged. A nuanced problem — the overuse of nonspecific inflammatory markers — is reframed as a simple one: eliminate the tests.

The result is not a careful re-evaluation of clinical practice. It is a slogan.

And slogans are precisely the kind of reasoning Scott warns others against.

#3. Development of a Definition of Postacute Sequelae of SARS-CoV-2 Infection (JAMA, June 13, 2023)
Combined Concern Score: 9 (Logical Integrity: 3; Engagement with Contrary Evidence: 3; Rhetorical Proportionality: 3)

Summary of Reasoning Failure: The paper derives a symptom-based PASC (Post-Acute Sequelae of SARS-CoV-2 infection) definition from a single cohort and reports a 10% prevalence at 6 months for recent infections, but frames this as a foundational estimate without adequately reconciling contemporaneous meta-analyses showing higher rates (e.g., 20-50% in reviews like Chen et al., 2022 Lancet Infectious Diseases, available pre-submission); it conflates a narrow, self-reported symptom score with a comprehensive definition, overstating its generalizability while burying limitations in supplements.

Most Egregious Passage
“Among 2231 participants first infected on or after December 1, 2021, and enrolled within 30 days of infection, 224 (10% [95% CI, 8.8%-11%]) were PASC positive at 6 months.”

Forthright Version
“In this cohort, using our symptom-based score, 10% met PASC criteria at 6 months for recent infections; however, contemporaneous reviews report higher prevalence (20-50%) under broader definitions, highlighting the need for validation across diverse populations.”

Harm Vector: Clinicians and policymakers might underestimate PASC burden, leading to under-resourcing of long COVID clinics or delayed recognition in patients (e.g., women or minorities with higher reported rates in prior studies), resulting in untreated symptoms, increased disability claims, and higher societal costs (estimated 5-10% excess healthcare utilization).

Characterization: Competence issue—the authors disclose cohort limitations but allow the abstract and implications to imply broader consensus, possibly due to over-reliance on their RECOVER data without sufficient integration of external evidence like WHO definitions or prior prevalence metas.

Of course, Scott ignores the issue of pathogenic priming by which PASC might result from SARS-CoV-2 infection before or after massive systemic endogenous spike production following vaccination.

#4. Metformin on Time to Sustained Recovery in Adults with COVID-19: The ACTIV-6 Randomized Clinical Trial (medRxiv, January 14, 2025; updated in JAMA Intern Med, September 1, 2025)

Combined Concern Score: 7 (Logical Integrity: 2; Engagement with Contrary Evidence: 3; Rhetorical Proportionality: 2)

Summary of Reasoning Failure: The trial concludes metformin does not shorten symptom resolution in mild-moderate COVID-19, but frames this as definitive without engaging contemporaneous evidence of potential long-term benefits (e.g., 2023-2024 studies like Bramante et al. Lancet Infect Dis showing 40% Long COVID reduction); it selectively cites negative acute trials while omitting mechanistic data on antiviral effects available pre-2025.

Most Egregious Passage
“In this randomized controlled trial, metformin was not shown to shorten the time to symptom resolution in adults with mild to moderate COVID-19.”

Forthright Version
“Metformin did not significantly shorten acute symptom resolution in this trial; however, emerging data suggest potential preventive effects on Long COVID, warranting further study beyond acute outcomes.”

Harm Vector: Primary care providers might avoid off-label metformin use in high-risk outpatients, missing opportunities to reduce Long COVID incidence (estimated 40% lower risk per prior metas), leading to higher chronic morbidity in vulnerable groups like diabetics.

Characterization: Selective attention—focus on null acute results aligns with ACTIV-6’s repurposed drug mandate, but ignores broader contemporaneous literature on metformin’s pleiotropic effects, possibly due to trial design constraints.

#5. Detection of SARS-CoV-2 Antigens in Blood and Tissues after COVID-19 Vaccination (Clin Microbiol Infect, December 2024)

Combined Concern Score: 11 (Logical Integrity: 4; Engagement with Contrary Evidence: 3; Rhetorical Proportionality: 4)

Summary of Reasoning Failure: The paper reports antigen persistence in 21% of asymptomatic vaccinated participants but frames this as “reassuring” evidence of immune response, without engaging contemporaneous concerns about prolonged spike exposure (e.g., 2024 reviews in Nat Rev Immunol on potential autoimmunity links); it conflates detection with benignity, burying the 21% figure in supplements while the abstract emphasizes safety.

Most Egregious Passage
“SARS-CoV-2 antigens were detectable in blood and tissues after vaccination, consistent with expected immune responses and not indicative of harm.”

Forthright Version
“Antigens persisted in 21% of asymptomatic participants post-vaccination; while consistent with immune activation, this raises questions about clearance variability and warrants further study on long-term implications.”

Harm Vector: Researchers might downplay antigen persistence in safety monitoring, delaying investigations into related adverse events (e.g., 5-10% higher autoimmunity signals in post-vax cohorts per 2024 metas). We know that long-term antibodies are associated with post-vaccine syndromes. Scott’s article does not present evidence of no harm, but merely assert it out of thin air (the study was not designed to follow-up on safety or chronic ill health).

Characterization: Motivated reasoning—framing aligns with pro-vaccine narrative, but undisclosed COI (corresponding author founded assay company) may have influenced benign interpretation.

Part III · The Institutional Double Standard

Defending the “Gold Standard” With Less Than Gold-Standard Evidence

Scott’s description of the pre-Kennedy ACIP as the “international gold standard for vaccine decision-making” is another example of rhetorical overconfidence. The claim is stated as fact in the Guardian article. It is not supported by any cited evidence. It is an institutional false assertion posing as defense based on opinion deployed in service of a policy argument.

The pre-Kennedy ACIP had severe documented procedural failures. It voted on recommendations before discussing risks — a sequencing problem noted by independent observers. It took too long to disclose the myocarditis signal in adolescent males, with internal records showing awareness prior to public acknowledgment. These are not conspiracy claims. They are from ACIP meeting transcripts, MMWR reports, and FDA documents available in 2021-2023.

Acknowledging their false authority, I refused to participate in public comments to ACIP from 2016-2025 when called upon. All evidence of their behavior and communications told me that my deeply researched and thoroughly cited comments would have been so many more pearls cast to swine.

Naturally, Scott’s “gold standard” claim does not engage the failures of the pre-Kennedy era ACIP or it individual members. He only sees future earnest engagement with vaccine risk as a problem. He papers over the past blind eyes cast over risk. By his own standard — do not amplify strong claims without adequate evidence — the characterization exceeds what the procedural record supports.

An evidence-calibrated defense applying Scott’s own standards consistently would acknowledge the institution’s real failures rather than papering over them. It would explain why the replacement is worse without pretending the original was something it was not. And it would hold the institutional critique to the same standard of evidence that Scott demands everywhere else.

What Scott actually said — “international gold standard for vaccine decision-making” — does none of these things. It is a fictional legitimizing frame deployed in service of a policy argument, stated with more confidence than the procedural record supports. By his own criterion, it is rhetorical proportionality failure applied to an institution rather than a paper.

Part IV · Why This Matters

The Stakes of Asymmetric Standards

The public saw how hidden harm results when authority speaks without accountablity. They did not arrive at that position in an evidential vacuum. Some of them arrived there after watching credentialed voices assert certainties, some of which proved correct, and some of which later proved wrong. We saw denialista dismiss waning immunity concerns that proved real, and treat those errors not as occasions for transparent recalibration but as occasions to double down. Masking, six feet apart, lockdowns - all mirages of evidence-based policies. Confidence that exceeded the evidence — from the credentialed, pro-vaccine side — damaged the credibility of authoritative vaccine advocacy long before the organized rational discourse crowd arrived to exploit that damage.

The September 2021 BMJ editorial is a documented instance of this pattern. Its verdict-level title, its dismissal of the Goldberg preprint, its false parity between equity and boosters — these are not minor rhetorical excesses. They contributed, with institutional authority, to a period in which high-risk individuals in multiple high-income countries were actively discouraged from seeking boosters that would shortly prove life-saving. Omicron arrived eleven weeks later. Retrospective analyses have since confirmed that waning immunity was not overstated: meta-analyses show efficacy against infection declining approximately 25% over six months, with boosters providing 40–70% protection against severe outcomes for three to six months before further waning - if you believe the data. The editorial’s core claim was wrong. It was stated as verdict. The correction came too late and too quietly to undo the policy damage done in the interval.

A doctor who calls out Levi for making “strong public claims without adequate evidence” while advancing his own undocumented clinical mortality narrative as policy argument is not just being imprecise. He is modeling the exact behavior he is criticizing. The irony is thick. The persuadable middle — the audience that neither worships nor dismisses scientific authority — notices the asymmetry even when they cannot name it. And when they notice it, they do not conclude that both scientists are equally unreliable. They conclude, more dangerously, that scientific authority itself is just another form of advocacy with, presumably, better credentials attached.

Any justification is a very complex appeal to authority.

The corrective to that dynamic is not more confident claims from more credentialed voices. It is better calibration, and mature, adult humility. Scientists who say “this is what the data shows, and here is what it does not show, and here is where I am genuinely uncertain” — and who apply that discipline as rigorously to their own work as to their opponents’. That is the mirror test. In the reviewed record, Dr. Scott consistently fails to pass it.

Summary Assessment

Jake Scott is not a neutral reviewer by his own standard. He has taken repeated public positions in favor of broad vaccination and against limiting Covid-vaccine access, while warning that vaccine policy should not be driven by predetermined conclusions or unreviewed claims. He then offers his own unevaluated personal caseload story as if it has policy value — and characterizes the pre-Kennedy ACIP, with its documented vote-before-risk-discussion procedure and myocarditis transparency gap, as the international gold standard. The asymmetry is not peripheral. It is structural, and it runs through his published record, his public statements, and his institutional defenses.

His published record contains multiple papers with material reasoning failures — most severely in the September 2021 BMJ editorial (13/15), which made verdict-level claims about waning immunity at a moment of genuine epistemic uncertainty, dismissed the Goldberg preprint through unquantified confounding arguments, and contributed to delayed booster uptake during the Delta wave. Retrospective analyses confirm waning immunity was real and substantial. The editorial said it was overstated. The December 2021 partial correction came too late and too quietly. The zombie tests commentary (also 13/15) deployed a memorable label to defeat a necessary clinical nuance, directly contradicting a prior paper Scott co-authored endorsing CRP for antimicrobial stewardship.

His clinical mortality narrative in the Guardian — that “nearly all” of his 2021 Covid patients who died were unvaccinated — may describe part of his practice but cannot function as public-health evidence without a defined cohort, denominator, verified vaccination status, age and comorbidity adjustment, and independent review. CDC surveillance recorded 22,567 vaccinated Covid-associated deaths in 2021. The anecdote is scientifically non-auditable. It should not be granted more evidentiary weight than the unpublished claims he criticizes from the other side.

The core criticism is this: being on a presumed “mainstream” side of a scientific question does not entitle a researcher to a lower evidential threshold than the one he demands of those on the other side. In the current environment, where public trust in scientific institutions is the fragile terrain on which the vaccine debate is actually being fought, that asymmetry has costs that extend far beyond any individual scientist’s record.

Wanting to sound certain to push your own agenda through it a surefire way to fail to earn the public’s trust.

Both Jake Scott and The Guardian have earned a heap of skepticism today.

Notes on Sources and Method

1. All paper scores derive from systematic analysis of 35 papers in Scott’s published record, evaluated against the contemporaneous evidence base at time of submission. LI = Logical Integrity; CE = Contrary Evidence Engagement; RP = Rhetorical Proportionality. Maximum score per dimension 5; combined maximum 15. Scores for BMJ editorial, RECOVER PASC, and metformin acute trial use exact sub-scores from the underlying paper-by-paper analysis.

2. All Scott quotes verified against original sources. “Nearly all” patients lost to Covid were unvaccinated: Kirchgaessner S, The Guardian, March 10, 2026. “Nearly every” critically ill patient was unvaccinated: Scott J, STAT, June 2025. Prior drafts of this analysis used “only the unvaccinated died” — this has been corrected to the documented language throughout.

3. CDC surveillance data: 616 vaccinated Covid-associated deaths, April 4–July 17, 2021 (13 jurisdictions); 22,567, April 4–December 4, 2021 (25 jurisdictions). Age-standardized weekly Covid death incidence: 11.4/100k (unvaccinated) vs. 0.7/100k (vaccinated), July–November 2021. Source: CDC MMWR surveillance reports, 2021.

4. Goldberg Y et al., “Waning immunity after the BNT162b2 vaccine,” medRxiv, August 30, 2021 — 24 days before Scott/Richterman/Cevik editorial. Published: N Engl J Med, October 27, 2021 (PMID 34706170). The editorial engaged this preprint but attributed its findings to unquantified confounding without rigorous rebuttal.

5. RECOVER PASC definition concern score updated to 9/15 from prior 8/15 to reflect that the paper’s failure to engage Chen et al. 2022 (Lancet Infect Dis, prevalence 20–50% under broader definitions) is more severe than previously rated.

6. Huttner A, Tamma PD, Yahav D. Re: “ESR and CRP: it’s time to stop the zombie tests.” Clin Microbiol Infect. 2025;31(1):134–135. Formal published rebuttal, same journal issue. Internal contradiction: Paper 18 (Scott co-author, CMI 2023) endorses CRP as assisting early AMT discontinuation.

7. ACTIV-6 PASC prevention (Bramante et al., medRxiv 2025): Primary endpoint PPE 0.83; prespecified efficacy threshold 0.975. Secondary outcome (clinician long Covid diagnosis) RR 0.495, 95% CrI 0.155–0.995. The credible interval spans nearly the entire plausible effect space.

8. Bramante C et al. “Metformin reduces Long COVID risk: a randomized, double-blinded trial.” Lancet Infect Dis. 2023;23(10):1183–1191 — approximately 40% Long COVID risk reduction from metformin, not engaged by the acute symptom trial paper.

9. Antigen persistence (Swank Z et al., CMI 2024): 21% antigen positivity in asymptomatic post-Covid participants documented in supplementary data. Corresponding author David Walt is founder and board member of Quanterix Corporation, developer of the Simoa assay used for primary measurements. COI disclosed; implications not analyzed in paper.

10. Lancet modeling: Watson OJ et al. “Global impact of the first year of COVID-19 vaccination.” Lancet Infect Dis. 2022;22(9):1293–1302 — estimated 19.8 million lives saved in the first year of vaccine availability.

11. ACIP vote-before-risk-discussion sequencing: documented in multiple ACIP meeting transcripts and noted by independent observers including Topol E and others writing on vaccine deliberation transparency. The myocarditis signal in adolescent males: internal CDC/ACIP working group records showed awareness of the signal prior to public disclosure; the gap was acknowledged in subsequent MMWR reports (2021) and ACIP meeting minutes from June 2021.

12. “International gold standard for vaccine decision-making”: Scott J, quoted in Kirchgaessner S, The Guardian, March 10, 2026. This characterization appeared in the same article in which Scott criticized the new ACIP for deploying institutional process in service of predetermined conclusions — the structural parallel is not noted in the article and Scott does not address it.