Discontinued

Beardsley is among the more prominent *proponents* of neuromechanical matching, not a sceptic of it. The passage praised him for a stance he does not hold — and did so while making a point about intellectual honesty, which makes it worse rather than better.

Now states the principle plainly, notes that Beardsley is a leading proponent, and flags it as genuinely contested with thin direct human evidence — so the reader knows where this site's reliance on the strict size principle is load-bearing.

Mandsager et al. (JAMA Network Open, 2018) reported *categorical* hazard ratios between fitness groups — most famously elite versus low fitness at around 0.20 — not a per-MET slope. The per-unit figure of roughly 13% per MET comes from Kodama et al.'s meta-analysis (JAMA, 2009). Two real results, merged into one citation that belonged to neither.

Both dossiers now separate the two explicitly. The conclusion — cardiorespiratory fitness is the largest single modifiable mortality lever, with no observed ceiling — was never in doubt; the citation was.

Matthews et al. (Diabetologia, 2009) — the primary source for muscle as a BDNF source — found muscle-derived BDNF is *not* released into the circulation. It acts locally, driving fat oxidation via AMPK. The circulating BDNF that rises with exercise comes largely from brain and platelets. The muscle-brain axis is real; this specific arrow in it is proposed rather than demonstrated, and the page asserted it as mechanism.

No meta-analysis reports that. The published dose-response meta-regressions (Schoenfeld, Ogborn & Krieger and later work) show hypertrophy rising with weekly sets along a decelerating curve — sharply diminishing returns, which is a weaker and different claim. The stronger 'first set does most of the work' reading is Beardsley's interpretation of that curve, not a pooled finding.

Now presents the shape of the curve as established and the magnitude as contested. The practical recommendation — low volume per session, higher frequency — survives, but on a narrower evidential base than the page originally claimed.

This is the textbook explanation and it has been specifically tested and not supported (Chalmers, Sports Biomechanics, 2004): EMG during the stretch phase does not show the predicted inhibition, and any reflex effect decays far faster than the range gain lasts. Worse, the page put this two paragraphs after correctly explaining that flexibility gains are a stretch-tolerance phenomenon — so it contradicted itself and reintroduced the model it had just displaced.

Energy loss in a tendon is hysteresis, and hysteresis and stiffness are independent properties. Human tendon returns roughly 90–93% of stored energy regardless of how stiff it is. What stiffness actually governs is timing and transmission — a stiffer tendon transmits force to the skeleton with less delay and less length change, which is what matters when ground contact lasts under 200 ms. The conclusion (train tendon stiffness for jumping) was right; the physics under it was wrong.

Swedroe argues *for* the passive and factor-based conclusion, not against it — he is one of its most prominent long-standing advocates. Presenting him as the opposition to the index-fund finding misdescribed a real person's public position in a section whose entire purpose is source accuracy.

The mouse-transfer result *is* the 2023 Nature paper, not a separate later follow-up. The page manufactured a phantom 2026 study and then — with unearned confidence in its own scepticism — flagged the phantom as unverifiable. Being appropriately cautious about a paper that does not exist is not rigour.

The lesson worth keeping: a hedge is not a substitute for checking. Flagging uncertainty about a fabricated citation looks like discipline and is the opposite.

Pelland et al. (Sports Medicine, 2026) pooled 67 studies and 2,058 participants with a novel direct/indirect set classification. Volume raises hypertrophy with a 100% posterior probability of a positive slope, showing diminishing returns. But frequency, for hypertrophy specifically, had a posterior probability below 100% and was described as compatible with negligible effects. Frequency clearly improves *strength*. The claim that it independently improves *growth* at matched volume is weaker than this site asserted.

The every-other-day structure survives on different grounds — session quality, recovery status, and the plain fact that a short session gets done — but not on 'frequency beats volume for growth'. The honest ranking is: total volume first, proximity to failure second, frequency as a distant third that mostly buys you a better way to distribute the volume.

Wolf, Androulakis-Korakakis, Roberts, Plotkin, Franchi, Contreras, Henselmans, Larsen & Schoenfeld's 2025 systematic review states plainly that no study has yet attempted to estimate serial sarcomere number changes from long- versus short-muscle-length training, and that all but one used linear extrapolation to estimate fascicle length — a method of questionable validity. The growth advantage at long lengths is reasonably supported. The specific explanation for it is not measured in humans at all.

A recurring pattern on this page: the finding was right and the mechanism attached to it was decoration. Training at long lengths still works. Nobody yet knows why at the sarcomere level.

Sleep target was adult-calibrated

Optimal was given as 7.5–8.5 h and elite as 8–9 h. The AASM/AAP consensus for ages 13–18 is 8–10 h — so the page's 'optimal' floor sat below the guideline floor, and its 'score to beat' was merely the ordinary paediatric recommendation. Now age-gated explicitly.

Body fat target was a RED-S risk

'8–12% held year-round' was listed as the elite target with no age qualification. Sustained low body fat in a developing male is the defining feature of relative energy deficiency in sport — it suppresses the HPG axis, impairs bone mineral accrual during the window when peak bone mass is set, and degrades exactly the training adaptations the rest of the protocol is chasing. Now reads 12–18% under 20, with sub-12% year-round explicitly not a target.

Oral minoxidil had no counterweight

It was the one row in the prescription table a seventeen-year-old could realistically obtain and plausibly want, and the only row whose evidence cell was pure endorsement. Now carries the dose-dependent facial and body hypertrichosis, fluid retention, tachycardia, postural hypotension, and the cardiovascular screening requirement.

Doses that exceeded their own ceilings

Acarbose was listed at 400 mg/day against an FDA-labelled maximum of 300 mg; magnesium's upper bound exceeded the 350 mg supplemental Tolerable Upper Intake Level; EPA/DHA at 2–3 g was filed under 'boring and they work' without noting it is a pharmacological dose; GlyNAC's per-kg framing concealed that it means roughly 7 g of each daily. All now carry the ceiling alongside the number.

The pattern across all four: the numbers were correct for the person the reference protocol was built around — a supervised 47-year-old — and were reproduced without re-deriving them for the person actually reading the page. That is the specific failure mode of copying any protocol, and this site committed it while warning against it.

This was the mainstream consensus position and it is not indefensible — the dose-response literature genuinely supports higher weekly set counts, particularly in advanced trainees, and the low-volume camp has not cleanly won. What replaced it (every other day, 2 sets of 5–8, each muscle ~3× weekly) follows from the stimulating-rep model: only fully-recruited, slow-enough reps count, later sets in a session degrade both requirements, and low per-session volume is what buys the frequency.

Retired for a reason, not because it was wrong. If the new structure stops producing progress, this is what to go back to — which is the entire point of keeping it written down.

Accurate as a target and useless as a protocol, because it described numbers rather than food. It was also quietly false in one respect — 'plants and fibre forward' did not describe what was actually being eaten. The replacement names the actual four items, does the leucine and micronutrient maths on them, and states the real gaps (fibre diversity, folate, K1, long-chain omega-3, saturated fat load) instead of implying they are covered.

The lesson generalises: a protocol written as targets rather than as behaviour cannot be audited, because there is nothing to check it against.

The first sentence repeated the popular claim that Zone 2 is the driver of mitochondrial biogenesis. It is not: Zone 2 produces only modest AMP accumulation and therefore modest AMPK/PGC-1α signalling, and the adaptations concentrate above roughly 65% of peak work rate. Zone 2 keeps its real value — capillarisation, fat oxidation, recoverable volume that does not tax lifting — but as the base rather than the stimulus. The second sentence also confused the intervals with the *quality*: VO₂max is the predictor; intervals are one way to raise it.

The content is unchanged in substance and in several places deeper. What changed is the frame: naming a protocol after a person makes the reader evaluate the person, and every subsequent judgement gets routed through whether they like him. De-naming it forces the only question that matters — is this lever supported, and by what.

Five of these seven had a beautiful, drawable mechanism. That is the recurring lesson of this entire page: a mechanism you can sketch on a whiteboard is a reason to run the trial, never a substitute for having run it.

The retractions above came from an adversarial multi-agent review run against the dossiers: four independent expert lenses (exercise physiology, clinical medicine, research integrity, philosophy and relationship science), each reading the source files directly and instructed to find errors rather than to approve. Every finding was then handed to a separate verifier prompted to *refute* it and to default to rejecting, so that plausible-sounding corrections did not survive on confidence alone.

• 37 agents run; 84 findings raised; 12 rated critical.
• Roughly 20 verifier agents terminated early on a session limit, so the review is thorough but not exhaustive — an unknown number of real errors remain unfound, and saying so is part of the point.
• Every critical and every safety-relevant finding was applied. The remaining minor findings are recorded and queued rather than discarded.
• The verify step earned its cost: several confidently-argued 'errors' were themselves wrong, including a claim that the typical-adolescent-sleep figure was understated when it was accurate.

The rule this page runs on: nothing gets deleted, only moved here. If a claim on this site changes, the old wording and the reason are preserved, because a protocol whose history is invisible cannot be audited — and an unauditable protocol is just a set of confident assertions with better typography.