The Ultra-Processed Food Debate: Science, Bias, & Truth
What this video covers
A Metabolic Mind discussion from September 2025 with host Dr. Bret Scher, Taubes, and Dr. David Ludwig, centered on two publications: Ludwig’s New England Journal of Medicine article criticizing the “ultra-processed food” category, and a Taubes Substack post dissecting a British trial published in Nature Medicine. The conversation becomes a debate about confirmation bias in nutrition science — including a pointed disagreement between Taubes and Ludwig themselves.
Key points
- Ludwig argues the NOVA definition of ultra-processed food, created by a Brazilian team in 2009, lumps benign and even beneficial processed foods into one “bad” category, and that this imprecision in policy and research can cause more harm than good.
- Taubes argues “ultra-processed food” is just a multi-syllabic term for junk food, and that trials feeding people chips, soda, and candy against salmon and kale prove nothing about processing itself — only that junk food is junk.
- Taubes dissects the eight-week British crossover trial in which both the ultra-processed and minimally processed groups lost weight; he argues the result refuted the researchers’ own hypothesis — drawn from Kevin Hall’s NIH study — yet was reported as confirming it.
- Taubes criticizes a Nature Medicine letter by Marion Nestel headlined “Ultra-processed diets promote excess calorie consumption” about a trial in which the ultra-processed group lost weight, using it to illustrate how a small circle of “usual suspects” shapes nutrition coverage.
- Taubes revives the phrase “Bing Crosby epidemiology” — accentuate the positive, eliminate the negative — for what he calls a field that only sees data confirming its beliefs, adding: “I’m proposing that they don’t know how to do science. And I do.”
- Ludwig pushes back on Taubes, calling that position “a little too facile” and urging their own camp to acknowledge its blinders; Taubes responds that his books hedge every claim as hypothesis, joking Good Calories, Bad Calories would have been 30 pages shorter without the qualifiers.