September 26, 2026 · We got it wrong
- What it said
- The page answered how much protein a plant-based lifter needs with about 1.6 g/kg a day, credited to 43 randomized trials, and said the source of protein stops mattering once enough is eaten ('the protein source debate is a dose debate in disguise'). It said 2.2 g/kg (1 g per pound) was 37% above what the evidence supports; that every subgroup (younger, older, men, women, trained, untrained) landed in the same place; that the 43 trials showed no relationship between the short-term muscle-building signal and long-term gains; and that the result for women was 'literally zero'. It quoted the review with a sentence the review does not contain, named the other plant proteins as pea, rice, wheat and hemp, called a non-randomised study an RCT and a parallel-group trial a crossover, gave the wrong review registration number and publication-bias test result, and said soy had no effect on testosterone, free testosterone or estradiol at any normal intake. It cited the 2025 review of short-term muscle-building studies (Mendes et al.) under a title that review does not have. After the rewrite went live on 26 September 2026, an independent second check against the same papers found more: the page called the five non-soy trials single proteins, although one gave oat and potato protein together; called the seven diet trials whole plant-based diets, although only one was a vegan diet; said blending rests on one short-term study, although the Mendes review lists several blend studies; said the daily target and the blending idea apply the same way to women, although both studies behind them tested only men; overstated what the blend trial showed about amino acids; called its 12-week study 'the one training study that matched intake', although every trial in the review matched protein; said in its audio that the non-soy gap 'was not about eating less protein', which the review does not show; gave the Mendes review's low certainty as due to risk of bias, where the review raised no concerns; and, in its evidence table, called wheat protein high in leucine and said the blend's amino acid profile beat that of any single plant protein, where the blend trial credits corn with the leucine and compares the blend only with its own three proteins.
- What it says now
- The review (Reid-McCann et al. 2025) matched protein between the groups in every trial and did not test amounts. Soy protein matched milk protein for muscle mass across 17 trials, with no difference by age or sex; other plant proteins (rice, chia, oat, potato; 5 trials) and plant-based diets (7 trials) gave less muscle than animal protein, and overall the gap was clearest in younger adults and in trials with resistance training. The page now says no trial here sets a separate target for plant-based lifters: 1.6 g/kg a day is the level one 12-week, non-randomised study of 38 untrained young men set for both groups, and the vegans on soy kept pace; it does not show whether more helps. The page quotes the review's own sentence, gives the women's result (no difference) and the men's (a trend towards animal protein that was not statistically significant), says the review cites earlier work on the short-term signal rather than testing it, and reports the testosterone review's estradiol result (no change overall, a rise in shorter trials and in trials without exercise). The Mendes review is cited under its own title, "Effects of plant- versus animal-based proteins on muscle protein synthesis: A systematic review with meta-analysis". The page also says the blend trial measured five hours after one drink, that the 1.6 g/kg study and the blend trial included only men, and that the Mendes review rated its certainty low for inconsistency and imprecision. Its evidence table now names corn as the source of the blend's leucine and methionine and pea as the source of its lysine, compares the blend with its own three proteins, and notes that the blend's lysine and methionine stayed below milk protein.
- Why
- The page was written from our synthesis notes, and its checks compared it with those notes rather than with the papers. The 1.6 g/kg figure comes from Hevia-Larraín et al. 2021, which set it for both groups as 'optimal protein intake', citing Morton et al. 2018; Morton's own estimate runs up to 2.2 g/kg and says about 2.2 g/kg 'may be prudent' for those seeking maximum gains, so '37% too high' was wrong as well. The quoted sentence was composed, not copied. A cross-page check while verifying a new Short on soy milk found the quote; reading the whole page against Reid-McCann 2025, Hevia-Larraín 2021, Mendes 2025, Pinckaers 2022 and Rajaie 2025 found the rest. The page's audio repeated these statements and was regenerated. Independent checks of the rewritten page against the same papers (the correction gate added on 26 September) found the later errors, and they were fixed before this record was published.
- Built on it
1 guide · 4 shorts · 7 recipes
- The paper
- Reid-McCann RJ, Brennan SF, Ward NA, Logan D, McKinley MC, McEvoy CT (2025). Effect of Plant Versus Animal Protein on Muscle Mass, Strength, Physical Performance, and Sarcopenia: A Systematic Review and Meta-analysis of Randomized Controlled Trials. Nutrition Reviews 83(7):e1581–e1603Read the paper
- Found by
- Our own reading of the whole page against its paper.