What happened when the 30 blank verdicts were filled in
Private deliverable. Enter the hub password to continue.
That's not right. Try again.
What happened when the 30 blank verdicts were filled in
7 of the 30 hold up. The blank reason turned out to be a signal, not an oversight.
The short version
The testosterone podcast audit had 30 claims published with a verdict and a blank explanation. Those blanks are now filled, and the audit page no longer has an empty cell anywhere.
Filling them turned up something worth more than the sentences. Only 7 of the 30 had evidence that both addressed the claim and carried the verdict assigned to it. For the other 23, the verdict itself is in question, so they say so on the page rather than carrying a confident explanation.
The blank was a signal. The original checker left the reason empty roughly where it could not justify itself. Filling all 30 with fluent explanations would have converted a visible gap into invisible false confidence, which is the failure this audit exists to catch.
How they were checked
Two passes, on different questions, neither shown the other's answer.
The first was asked to write the explanation from the evidence already retrieved. It produced a usable sentence for all 30 and reported a problem with none of them, which is the expected failure: a model asked to explain a verdict will find a way to explain it.
The second was asked only whether the passage is about the claim at all, and what verdict the passage actually carries. It never saw the first pass's sentence. Disagreement between them is what produced the 23.
Nothing was re-verified against the literature and no new sources were consulted. Both passes read only the evidence the original run had already retrieved.
The 23 where the verdict is in question
| What went wrong | Count |
|---|---|
| passage does not settle it | 12 |
| wrong passage | 4 |
| evidence supports the claim | 4 |
| contradicted, not overstated | 3 |
"Wrong passage" means the retrieved text is about a different subject entirely. "Evidence supports the claim" means the podcast may have been right and the verdict wrong. "Contradicted, not overstated" means the verdict is too generous: the passage says the opposite of the claim rather than a weaker version of it.
| Episode | At | Verdict given | Claim | What the second read found |
|---|---|---|---|---|
| Hormone Therapy and Endometrial Health in Postmenopausal Women | 11:09 | OVERSTATED | Unopposed estrogen use increases endometrial cancer risk from about 1.8 times higher with any use to over 3.5 times higher after five years. | The passage discusses unopposed estrogen and endometrial cancer risk but gives different magnitude (hyperplasia risk 2-4x) and no time frame, so it does not address the specific claim numbers. |
| Hormone therapy reduces mortality after sixty-five | 12:09 | OVERSTATED | Non-oral routes are vastly superior for long-term health. | The passage states that oral and non-oral routes have comparable efficacy and safety, directly contradicting the claim that non-oral routes are vastly superior. |
| Hormone therapy reduces mortality after sixty-five | 15:35 | OVERSTATED | Transdermal delivery of human-identical estradiol bypasses liver inflammatory response, preventing heart failure, colon cancer, and neurodegeneration. | Passage only mentions bypassing hepatic metabolism and adverse hepatic effects, not the specific disease preventions (heart failure, colon cancer, neurodegeneration) claimed. |
| How Fat Turns Testosterone Into Estrogen | 17:01 | OVERSTATED | TRT significantly improves sexual desire, restores erectile function, increases lean muscle mass, and reverses insulin resistance. | The passage addresses improvements in sexual desire, erectile function, and lean body mass but omits any mention of insulin resistance, leaving a key part of the claim unjudgeable. |
| How Testosterone Protects Women From Alzheimer's | 15:56 | OVERSTATED | The APOE4 gene drastically increases baseline neuroinflammation. | The passage mentions higher neuroinflammation in older adults but does not reference the APOE4 gene, so it is unrelated to the claim. |
| How Testosterone Protects Women From Alzheimer's | 19:27 | OVERSTATED | Resistance exercise is the most accessible and effective lifestyle approach to modulate testosterone levels naturally. | The passage only states that both aerobic and resistance training can increase testosterone, not that resistance exercise is the most accessible and effective approach. |
| Menopausal Transition Increases Depression Risk in Later Life | 12:20 | OVERSTATED | The symptomatic menopausal transition seems to add its own independent layer of risk for developing depression later. | The passage addresses depression risk during the menopausal transition, not the claim that the transition adds independent risk for developing depression later. |
| Menopausal Transition Increases Depression Risk in Later Life | 15:26 | OVERSTATED | Symptomatic menopausal transition appears to be a factor associated with an increased likelihood of facing clinical depression later. | The passage discusses early menopause as a risk, not symptomatic menopausal transition, so it does not address the specific claim. |
| Precision Hormones and Nanotechnology for Depression | 6:54 | OVERSTATED | The stroke risk in the WHI study was caused by oral hormones passing through the liver and triggering clotting factors. | The passage directly states that oral estrogens increase clotting factors via hepatic metabolism, which matches the claim's mechanism for stroke risk. |
| Progesterone Acts as Literal Brain Armor | 16:26 | CONTRADICTED | The Protect the Third and Synapse trials failed to show clinical benefit. | The passage discusses memory benefits from sustained engagement, not the Protect the Third or Synapse trials. |
| Progesterone Acts as Literal Brain Armor | 17:46 | OVERSTATED | APOE4 is a major genetic risk factor for Alzheimer's disease and is associated with aggressive decline. | The passage discusses the APOE gene broadly, not the specific APOE4 allele, and omits the claim about aggressive decline entirely. |
| Testosterone Regulates Memory and Mood | 15:08 | OVERSTATED | In a subset of men with clinically low testosterone, supplementing testosterone alongside standard psychiatric care significantly dropped depression scores. | The passage states that evidence for TRT treating depression in men with testosterone deficiency is not convincing, directly contradicting the claim that it significantly dropped depression scores. |
| Testosterone Therapy for Women's Desire and Mood | 6:10 | OVERSTATED | Only about 50% of circulating testosterone comes directly from the ovaries and adrenal glands. | The passage indicates 25% from ovaries and 25% from adrenal glands, totaling 50% directly, which directly supports the claim. |
| Testosterone rebuilds skin to silence eczema | 12:38 | OVERSTATED | Filaggrin is arguably the most important structural protein in the human epidermis. | The passage notes filaggrin's crucial role in barrier function but does not compare it to other structural proteins or claim it is the most important. |
| Testosterone therapy for female sexual desire | 18:27 | OVERSTATED | Oral testosterone formulations are strongly discouraged due to adverse effects on lipid profiles and liver function. | The passage only discusses adverse lipid profiles and does not mention liver function, so it cannot fully evaluate the claim. |
| Testosterone, GABA, and Depression in Women | 10:10 | OVERSTATED | Less free testosterone was potentially associated with more severe depression symptoms. | The passage discusses brain activity (ACC and PCC), not depression symptoms, so it does not address the specific outcome in the claim. |
| The Dangerous Convenience of Testosterone Pellets | 12:38 | OVERSTATED | Observational studies report benefits beyond treating HSTD when physiological equivalence is achieved with testosterone pellets. | The passage addresses whether physiological doses are adequate, not the specific claim about observational studies reporting benefits beyond HSTD at physiological equivalence. |
| The Risks of Female Testosterone Pellets | 19:45 | OVERSTATED | Genetically higher testosterone drives visceral fat and diabetes in women. | The passage states that androgen excess (including testosterone) drives adipose dysfunction and metabolic deterioration in women, directly supporting the claim. |
| What testosterone does to your brain | 17:43 | OVERSTATED | Observational data shows a spectacular mood and motivation boost from hormone therapy in women. | The passage discusses RCT evidence on testosterone for menopausal symptoms, not observational data on mood and motivation boost from hormone therapy. |
| Why Your Hormone Blood Tests Lie | 14:15 | OVERSTATED | The ratio of standard testosterone to 11-KT in a peripheral blood draw can help determine if androgen excess is ovarian or adrenal. | The passage directly proposes the same ratio to help distinguish adrenal from ovarian androgen excess, matching the claim. |
| Why total testosterone is a meaningless metric | 5:52 | OVERSTATED | If androgen receptors do not move into the nucleus, muscle does not grow regardless of how much testosterone is outside the cell. | The passage describes the translocation of androgen receptors upon ligand binding but does not address the claim about muscle growth or the effect of receptor non-translocation. |
| Why women have zero testosterone options | 15:03 | OVERSTATED | Compounding pharmacies are held to strict standards for testing purity, potency, and sterility under CGMP. | The passage states these establishments are exempt from CGMP, directly contradicting the claim that they are held to strict standards under CGMP. |
| Why women need testosterone after menopause | 23:31 | OVERSTATED | Compounded pellets and intramuscular injections are known to cause dangerous, uncontrollable spikes in hormone levels. | The passage addresses only intramuscular injections, not compounded pellets, so it cannot judge the full claim. |
The 7 that hold up
Evidence addresses the claim directly and carries the verdict it was given. These now read on the audit page exactly as written here.
| Episode | At | Verdict | Claim | Reason, now published |
|---|---|---|---|---|
| How Testosterone Shapes Female Biology | 17:40 | CONTRADICTED | In the HSR mouse model, female mice overexpressing the androgen receptor in muscle tissue, when exposed to androgens, rapidly gained lean mass and grew larger glycolytic muscle fibers. | The podcast claims female mice rapidly gained lean mass and grew larger muscle fibers, but the passage states that testosterone treatment in the same HSAAR mouse model induced drastic declines in body mass and motor function, not growth. |
| How muscle strength saves your brain | 15:08 | OVERSTATED | The RDA of 0.8 grams of protein per kilogram is dangerously low for older adults. | The podcast asserts international guidelines are adamant that 0.8 g/kg is dangerously low, but the passage states that WHO and EFSA recommend 0.83 g/kg for all adults, with only some researchers arguing it might underestimate older adults' needs. |
| Testosterone for female sexual desire | 7:16 | OVERSTATED | Testosterone therapy is highly effective in postmenopausal women. | The passage states that research supports a moderate therapeutic benefit, not the high effectiveness claimed in the podcast. |
| Testosterone for female sexual desire | 21:47 | OVERSTATED | For postmenopausal HSDD, testosterone therapy is highly effective, guideline-backed, and powerful. | The podcast describes testosterone therapy as 'highly effective,' but the supporting evidence reports only a 'moderate therapeutic benefit,' overstating the actual effect. |
| Why Testosterone Needs Estrogen to Burn Fat | 14:19 | CONTRADICTED | Female testosterone levels peak between ages 20 and 25 and then slowly decline. | The passage states female testosterone peaks around age 30, while the podcast claims the peak is between 20 and 25. |
| Why women have more testosterone than estrogen | 19:46 | CONTRADICTED | Testosterone therapy must be guided by clinical symptoms and safety tolerability rather than lab targets. | The passage states that delaying a procedure is indicated when testosterone levels are supraphysiologic or hematocrit exceeds a specific lab target, which directly contradicts the podcast's claim that therapy should not be guided by lab targets. |
| Why women have zero testosterone options | 16:40 | CONTRADICTED | There is zero evidence testosterone administration promotes endometrial cancer. | The study cited in the passage found that testosterone increased the risk of endometrial cancer, directly contradicting the claim that there is zero evidence for this effect. |
What changed on disk
Every backfilled record carries why_source: backfill-2026-08-10, so a sentence
written today is never mistaken for the original verification run, and the 23
questioned rows also carry why_review: true.
The 46 verdict files were untracked, so nothing was backing up the whole claim-by-claim result and no change to it could be reviewed. They are in git now, committed before the backfill, so this edit is a diff against a known-good base.
build_report.py now refuses to build when a published verdict has no reason,
rather than rendering it as an empty cell. That is the bug that let 30 silent
blanks reach a live page.
What this does not tell you
These 30 are the audit's weakest rows by construction: they were selected because the original checker failed to produce a reason for them, so a high problem rate here says little about the other 1,351 claims. It is a reason to spot-check the rest, not to discount them.
The second read is another model's judgment on the same retrieved passage, not a fresh look at the literature. Where it says the evidence supports the claim, that means the retrieved passage supports it. Settling any of these properly needs the episode's actual source resolved, which the audit has never done.
Published to Annette's hub. Rebuilt from the source markdown, so edit the source and rerun rather than editing this page.