Chapter 27 — Gut Health and the Microbiome: What Sequencing Can Tell You, and What It Can't

🚪 Threshold concept: the microbiome is genuinely important, and almost nothing sold to "fix" it works yet. ⚠️ Both halves are required. Holding only the first makes you a customer; holding only the second makes you dismiss real biology. The whole chapter is an exercise in keeping them at the same time.

The Hook: Theo's PDF

Theo Vasquez paid £179 for a microbiome test. It arrived as a fourteen-page PDF.

Page 3 gave him a "Gut Health Score" of 64/100.

Page 7 told him his Akkermansia muciniphila was low and recommended pomegranate extract.

Page 9 listed forty-one foods to "increase" and twenty-three to "reduce." ⚠️ Among the foods to reduce were oats, apples and onions.

Page 11 said his profile was "associated with" a raised risk of several conditions.

He brought it in and asked the only question that matters:

"Is any of this real?"

⚠️ Here is the honest answer, and it has three parts:

1. ⚠️ The measurement is real. The sequencing did detect the organisms it says it detected, more or less, subject to the platform's method.

2. ⚠️ The interpretation is not validated. There is no established reference range for a "healthy" microbiome, no clinical validation of the score, and no trial showing that acting on the recommendations improves anything.

3. ⚠️ And the recommendations partly contradict the best-supported advice in this chapter. Oats, apples and onions are among the most fermentable-fibre-rich foods available — which is what §27.4 says the whole system runs on.

⚠️ Send the same stool sample to two different companies and you will frequently get materially different profiles and different recommendations, because the platforms use different sequencing methods, different reference databases and different analysis pipelines. Comparisons of this kind have been done and the agreement is poor.

⚠️ This is the chapter with the largest gap in the whole book between scientific excitement and clinical usefulness — and I want to be clear that the science is genuinely exciting.

The microbiome is one of the most interesting areas in biology. It is also the area where the distance between "we have discovered something remarkable" and "therefore you should buy this" is shortest, and where that distance is crossed most often.

⚠️ So this chapter is structured around a single distinction: what is established, and what is interesting. Both are real categories. They are not the same category, and almost every gut-health product on the market depends on you not noticing the difference.

🏃 Fast Track: §27.4 (SCFAs and why fibre keeps returning), §27.6 (⚠️ probiotics that work), §27.9 (🟢 fermented foods), §27.11 (⚠️ IBS — the most useful section). Thirty-five minutes.

🔬 Deep Dive: §27.2 (what sequencing can't tell you), §27.7 (⚠️ the probiotic findings that surprised everyone), §27.10 (FMT).

27.1 What the microbiome actually is

Trillions of microorganisms, overwhelmingly bacterial, overwhelmingly in the colon — plus archaea, fungi, and a very large population of viruses that infect the bacteria.

⚠️ A correction worth making, because the old figure is still everywhere: the claim that microbial cells outnumber human cells 10 to 1 was revised in 2016 to a ratio much closer to 1:1. The original figure came from an estimate that propagated for decades without being checked — ⚠️ which is Chapter 15's eight-glasses problem in a different field, and worth noticing.

What the community does:

⚠️ And the single most important structural fact: there is no defined "healthy microbiome." Inter-individual variation is enormous, healthy people differ from each other far more than most disease states differ from health, and no reference range exists.

27.2 What sequencing can and can't tell you

Two main methods, and the difference matters.

⚠️ And here is what NEITHER tells you:

1. ⚠️ What the community is actually DOING. DNA tells you what organisms and genes are present. It does not tell you which genes are being expressed, at what rate, or what metabolites are being produced. That requires metatranscriptomics or metabolomics, which consumer tests do not do.

2. ⚠️ Whether a finding is causal. Almost all human microbiome data is cross-sectional association. A taxon differing between people with and without a condition may be a cause, a consequence, or a marker of something else entirely — Chapter 2's reverse causation, at scale.

3. ⚠️ What your gut looks like. Stool sampling captures the luminal community in the distal colon. The mucosal community — the one in contact with your intestinal wall — differs, and so does the small intestine.

4. And what it will look like next week. ⚠️ Composition varies day to day with diet, transit time, medication and illness.

💡 ⚠️ The clearest way to hold this: sequencing gives you a photograph of who was in the room, taken at the door, on one day.

It does not tell you what they were doing, why they were there, whether it mattered, or who was in the other rooms.

Which is not a criticism of the science — it is a description of what the method measures, and the science knows this perfectly well. ⚠️ It is a criticism of what gets sold on top of it.

27.3 ⚠️ Consumer microbiome testing

🔬 Claim → Evidence → Verdict

The claim: "A microbiome test can tell you what to eat, and identify your health risks."

What's true: the sequencing works. The organisms detected are, broadly, there.

What isn't:

1. ⚠️ No validated reference range. There is no established "healthy" profile to compare against, so a "score" is a number derived from a proprietary comparison to an unvalidated standard.

2. ⚠️ Poor agreement between platforms. Different sequencing approaches, reference databases and bioinformatic pipelines produce different results from the same sample.

3. ⚠️ Recommendations are generated from associations, not from trials. No consumer microbiome test has demonstrated in a randomized trial that following its advice improves a clinical outcome.

4. ⚠️ And no regulatory framework applies. These are sold as wellness products, not as diagnostics — which is Chapter 16 §16.1's DSHEA structure, arriving in a laboratory.

📉 Evidence quality: The measurement is a real measurement. The interpretation is unvalidated.

Verdict: ❌ Not supported as a guide to what to eat or as a health risk assessment.

⚠️ And note the specific harm, which is not the £179: Theo was told to reduce oats, apples and onions. Acting on that would have reduced his fermentable fibre — which is the one intervention in this chapter with genuine evidence behind it.

⚠️ Why two labs disagree about the same stool

Worth understanding, because it explains the whole category rather than just one company.

💡 ⚠️ None of that is fraud. Every step is a defensible methodological choice, and researchers know about all of them — which is why serious studies standardize the whole pipeline and compare samples processed identically.

⚠️ What you cannot do is take an absolute number from one pipeline and interpret it as a fact about a person. Which is exactly what a consumer report does.

The measurement is relative. The report presents it as absolute.

27.4 Short-chain fatty acids, and why fibre keeps coming back

⚠️ This is the mechanistic core of the chapter, and it is why Chapter 11 keeps reappearing in every chapter since.

Bacteria in the colon ferment carbohydrate that human enzymes cannot digest — principally fermentable fibre and resistant starch — and produce short-chain fatty acids.

What SCFAs do, in decreasing order of confidence:

• ✅ Feed the colonic epithelium — butyrate is its preferred fuel
• 🟢 Support gut barrier function
• 🟢 Influence immune regulation, including regulatory T cell populations
• 🟢 Stimulate release of PYY and GLP-1 — ⚠️ the same satiety hormones as Chapters 3 and 24
• 🟡 Influence hepatic metabolism and appetite regulation

💡 ⚠️ This is the thread that runs through the entire book, and it's worth naming explicitly.

Chapter 3's digestion, Chapter 7's intact-versus-refined carbohydrate, Chapter 11's fibre, Chapter 22's ultra-processed food, Chapter 26's metabolic disease — and now Chapter 27's microbiome.

⚠️ They converge on the same recommendation from six different directions: eat more fermentable fibre from whole plant foods.

That is not a coincidence and it is not repetition. It is what a real finding looks like when you approach it from different fields — and it is the strongest structural argument in this book.

⚠️ And it explains why "feed your microbiome" products are mostly unnecessary: the substrate is food. Beans, oats, barley, onions, garlic, leeks, bananas, cooled potatoes and rice, and every other source in Chapter 11 §11.4.

27.4b What actually changes your microbiome, ranked

⚠️ The cross-cutting section this chapter needs, because the marketing implies the order is the reverse of what it is.

💡 ⚠️ Read rows 1 and 9 together.

The most powerful thing that will ever happen to your microbiome is a course of antibiotics — and the thing sold to you as microbiome care is at the bottom of the list.

⚠️ And row 5 is the one nobody mentions. Proton pump inhibitors and metformin produce measurable, reproducible shifts in gut microbial composition — which means a substantial fraction of people who buy a microbiome test are measuring their prescription, not their diet.

Walt Prosser is on both (Chapters 13 and 16). Any test he took would have found something, and the something would have been pharmacological.

⚠️ Two practical consequences:

1. ⚠️ Don't take antibiotics you don't need — which is a public health message about resistance and is also, separately, the most effective microbiome advice available.

2. ⚠️ And judge dietary change over months, not days. Short-term shifts are real and revert. The community that matters is the one your habitual diet has been selecting for over years.

27.4c Early life, briefly

⚠️ Chapter 25's material, because this is where the community is established and where the associations are strongest.

⚠️ The human milk oligosaccharide finding is the one I'd keep. Breast milk contains a large quantity of complex carbohydrate that the baby cannot digest at all. It is not there for the infant. It is there for the infant's bacteria — which is about as direct a piece of evidence as biology offers that this relationship is not incidental.

⚠️ And the caution: the associations between early-life microbiome disruption and later allergy, asthma, obesity and autoimmunity are real associations and remain substantially confounded. Caesarean birth, formula feeding and early antibiotics all cluster with other exposures. Chapter 25 §25.3's warning about loading responsibility onto mothers applies here with force.

27.5 Diversity: a metric that got ahead of itself

Higher microbial diversity is associated with health in many contexts, and it became a headline number.

⚠️ It is a summary statistic, not a goal, and treating it as a target produces errors.

Three reasons:

1. ⚠️ Some healthy states have LOW diversity. The breastfed infant gut is dominated by bifidobacteria and is low-diversity by design (Chapter 25).

2. ⚠️ Some disease states have HIGH diversity. Small intestinal bacterial overgrowth is a diversity problem in the wrong place.

3. And it says nothing about function. ⚠️ Two communities with identical diversity indices can do completely different things.

🔬 Verdict: 🟡 Unclear / it depends. ⚠️ Diversity is a useful population-level correlate and a poor individual target — and "increase your diversity" is advice with no defined endpoint, which Chapter 20 §20.13 identified as a structural problem.

27.6 Probiotics: the list that actually has evidence

⚠️ Shorter than the shelf suggests, and the first thing to understand is why it's a list of STRAINS rather than a list of products.

⚠️ PROBIOTIC EFFECTS ARE STRAIN-SPECIFIC. Evidence for one strain does not transfer to another strain of the same species, let alone to a different product.

A trial showing benefit for Limosilactobacillus reuteri DSM 17938 says nothing about a supermarket drink containing a different Lactobacillus. This is the single most important fact in §27.6 and it is almost never conveyed on packaging.

Where the evidence is genuinely reasonable:

⚠️ Two things on that table need their own paragraph.

⚠️ The paediatric gastroenteritis reversal. Probiotics for acute childhood diarrhoea were widely recommended on the basis of earlier trials. Then two large, well-conducted randomized trials published in the New England Journal of Medicine in 2018 found no benefit for L. rhamnosus GG in children with acute gastroenteritis. ⚠️ This is a clean example of a larger trial deflating an accepted finding — Chapter 2 §2.8's pattern, and the correct response is to update.

⚠️ And the neonatal safety note, which matters enormously. The NEC evidence is genuinely good. But in 2023 the FDA issued warnings about probiotic use in preterm infants following an infant death associated with a contaminated product. ⚠️ Chapter 16 §16.2's manufacturing-quality problem, in the most vulnerable population imaginable. This is a hospital decision with hospital-grade products, not a parental one.

27.7 ⚠️ What probiotics don't do — including two findings that surprised everyone

❌ or 🟠 for: "general gut health" · "immune support" · weight loss · mood in healthy people · and essentially every claim made on a supermarket shelf that doesn't name a strain and an indication.

And three mechanistic findings that changed how the field thinks:

1. ⚠️ Most probiotics don't colonize. They pass through. Effects, where they exist, generally require continued intake and stop when you stop — which is a legitimate way for something to work and is not what most people assume they're buying.

2. ⚠️ Colonization resistance is personalized. Work published in Cell in 2018 (Zmora and colleagues) found that some people's guts permitted probiotic colonization and others' resisted it — and that stool sampling did not reliably predict which.

3. ⚠️ And the one that genuinely surprised people: after antibiotics, probiotics DELAYED microbiome recovery. A companion study (Suez and colleagues) compared, after an antibiotic course, spontaneous recovery, probiotic supplementation, and autologous faecal transplant — reinfusing the person's own pre-antibiotic stool. ⚠️ The autologous transplant restored the original community fastest. Spontaneous recovery was intermediate. The probiotic group's return to baseline was SLOWEST.

⚠️ These are small mechanistic studies, not outcome trials, and I'd hold them at moderate confidence.

But they matter because they invert the intuitive model. "Take a probiotic after antibiotics to restore your gut" is the single most common piece of gut-health advice given, and the best mechanistic evidence available suggests it may do the opposite of what it intends.

⚠️ This is Chapter 19 §19.5 again: a clean mechanism, a confident recommendation, and a direct test that didn't go the expected way.

27.8 Prebiotics

Non-digestible compounds that selectively feed particular organisms — inulin, fructo-oligosaccharides (FOS), galacto-oligosaccharides (GOS), and resistant starch.

What they do: ⚠️ reliably increase bifidobacteria and produce SCFAs. That part is well demonstrated.

What's less clear: 🟡 whether that translates into clinical outcomes in healthy people.

⚠️ And a practical warning: prebiotic supplements commonly cause bloating, wind and discomfort, particularly at the doses sold — and they are frequently the FODMAPs that §27.11 is about.

🔬 Verdict: 🟢 that prebiotics change the microbiota as described; 🟡 that supplementing them improves clinical outcomes in healthy people. ⚠️ And Chapter 11 §11.6's position holds: the food versions do the same job, cost less, and come with everything else.

27.9 🟢 Fermented foods — the verdict this book has owed since Chapter 11

⚠️ Chapter 11 promised this and here it is.

🔬 Claim → Evidence → Verdict

The claim: "Regularly eating fermented foods improves microbiome diversity and reduces inflammation."

What the evidence shows: ⚠️ A randomized trial published in Cell in 2021 (Wastyk and colleagues, from Stanford — the Sonnenburg and Gardner groups) compared a high-fermented-food diet with a high-fibre diet over ten weeks in healthy adults.

The findings were not what most people, including the investigators, would have predicted:

	⚠️ High fermented food	High fibre
Microbiota diversity	⚠️ INCREASED	⚠️ Did not increase
Inflammatory markers	⚠️ DECREASED — across a substantial panel	Variable and personalized

⚠️ The fermented-food arm increased microbial diversity and reduced inflammatory markers. The high-fibre arm — over ten weeks, in people starting from a low-fibre baseline — did not increase diversity, and its inflammatory response depended on the individual's starting microbiome.

📉 Evidence quality: ⚠️ Randomized dietary intervention with mechanistic outcomes, modest sample, ten weeks, healthy adults. Not an outcome trial.

Verdict: 🟢 Probably true.

⚠️ Three qualifications I'd attach:

1. ⚠️ This does NOT mean fibre doesn't matter. Chapter 11's evidence is about cardiovascular disease, colorectal cancer, glycaemia and mortality — hard outcomes over decades. This trial measured diversity and inflammatory markers over ten weeks. ⚠️ Different questions, different timescales. The plausible interpretation is that ten weeks was too short for the fibre arm's community to adapt, not that fibre is unimportant.

2. ⚠️ "Fermented" is not one category. Many fermented foods contain no live microbes by the time you eat them.

⚠️ Kefir typically contains a more diverse microbial population than yoghurt. And "pickled" and "fermented" are not synonyms — most supermarket pickles are preserved in vinegar and were never fermented at all.

3. ⚠️ And watch the sodium. Kimchi and sauerkraut are salty. In someone managing blood pressure (Chapter 26), that's a real consideration — though the quantities involved are usually small.

⚠️ Practical translation: a serving or two a day of something with live cultures. Yoghurt, kefir, a spoonful of kimchi. It is cheap, it is food, and it has the best randomized dietary evidence in this chapter.

27.10 Faecal microbiota transplantation

⚠️ The intervention that works spectacularly for one condition and unimpressively for the others — which is itself informative.

✅ Recurrent Clostridioides difficile infection. ⚠️ FMT has produced cure rates in trials substantially exceeding standard antibiotic therapy, and it is established in clinical guidelines.

Why it works so well here is instructive: ⚠️ recurrent C. difficile is a disease of destroyed colonization resistance. The community has been wiped out by antibiotics and C. difficile has occupied the vacancy. Restoring a functioning community restores the resistance.

🟡 to ⚗️ for everything else. Ulcerative colitis has shown some signal in trials. IBS, metabolic syndrome, obesity, autism and a long list of other proposed indications have much weaker or absent evidence.

💡 ⚠️ The pattern is the useful part: FMT works where the problem IS the missing community.

Where the microbiome is one contributor among many to a complex disease, replacing it does much less — which should calibrate expectations for every microbiome-based therapy, including the ones not yet invented.

⚠️ And the safety point is severe. In 2019 the FDA reported serious infections, including a death, caused by transmission of a drug-resistant organism from inadequately screened donor material. Donor screening is extensive and it exists for this reason.

⚠️ DIY faecal transplant — which people do, using material from friends and family, following online instructions — is genuinely dangerous. It can transmit infections including drug-resistant organisms, hepatitis and HIV. There is no safe way to do this at home.

27.11 ⚠️ IBS and low-FODMAP: the most useful thing in this chapter

And the part most often done badly.

Irritable bowel syndrome is now framed as a disorder of gut–brain interaction — ⚠️ which replaced the older term "functional," and the change matters, because "functional" was frequently heard as "not real." IBS is real, common, and its symptoms are generated by genuine alterations in gut motility, visceral sensitivity and gut–brain signalling.

FODMAPs

⚠️ Fermentable Oligosaccharides, Disaccharides, Monosaccharides And Polyols — short-chain carbohydrates that are poorly absorbed, osmotically active, and rapidly fermented. The framework was developed at Monash University.

In a sensitive gut, they draw water into the lumen and ferment rapidly, producing distension, pain and altered bowel habit.

🔬 Verdict: ✅ Well supported that a low-FODMAP diet reduces symptoms in a substantial proportion of people with IBS — trials commonly report meaningful improvement in something like half to three-quarters of participants, which is a strong result for a dietary intervention.

⚠️ THE THREE PHASES — and the failure

Phase 1 — RESTRICTION. ⚠️ 2–6 weeks. Not longer. Phase 2 — SYSTEMATIC REINTRODUCTION. ⚠️ Testing each FODMAP group individually to identify which ones actually trigger symptoms, and at what dose. Phase 3 — PERSONALIZATION. A long-term diet restricting only what's necessary.

⚠️ The failure mode is that people do phase 1 and stop.

And it happens constantly — because phase 1 works, so there is no symptomatic reason to move on, and because phases 2 and 3 require guidance that many people never get.

⚠️ What staying in phase 1 costs:

⚠️ Chapter 26 §26.11b's pattern, exactly: a motivated, symptomatic person handed a restriction by someone who meant well — and the restriction becomes the harm.

Low-FODMAP is a diagnostic process, not a diet. Its purpose is to find out what your triggers are so you can eat everything else.

⚠️ It should be dietitian-led, and where that isn't available, the reintroduction phase is the part to insist on.

⚠️ What reintroduction actually involves, because "reintroduce foods" is not instructions:

Each FODMAP group is tested separately, using a food that contains that group and little else, at increasing doses over about three days, with a washout period between challenges and symptoms recorded daily. ⚠️ Lactose might be tested with milk; fructans with bread and separately with onion; galacto-oligosaccharides with chickpeas; excess fructose with honey; sorbitol with blackberries; mannitol with mushrooms.

⚠️ Two things this design is doing that people miss:

1. ⚠️ It tests groups separately because most people react to some and not others. Avoiding all seven when you react to two is five unnecessary restrictions.

2. ⚠️ And it tests DOSES, because most reactions are threshold effects rather than absolutes. Someone who reacts to two slices of bread may be entirely fine with one — which is a completely different life from "I can't eat wheat."

⚠️ And what should usually come first

Before low-FODMAP, standard first-line dietary advice helps many people and is much less restrictive: regular meals, not skipping · limiting caffeine, alcohol, fizzy drinks, fatty and spicy food · adjusting fibre TYPE — ⚠️ soluble fibre such as psyllium helps, while coarse insoluble bran can worsen symptoms (Chapter 11 §11.3) · reducing very large meals · and addressing constipation or diarrhoea specifically.

Also worth knowing:

🟢 Peppermint oil has reasonable evidence for IBS symptoms. ⚠️ 🟢 And gut-directed hypnotherapy and cognitive behavioural therapy have trial evidence comparable to dietary intervention. That is not a consolation prize and it is not "it's in your head" — it is a treatment that acts on the gut–brain axis, which is where the disorder lives.

⚠️ "Leaky gut" — the term, and what's underneath it

It comes up constantly and it needs separating into two things.

⚠️ Intestinal permeability is real and measurable. The epithelium is a regulated barrier, tight junctions can loosen, and increased permeability is documented in coeliac disease, inflammatory bowel disease, severe illness and after heavy alcohol exposure. This is established physiology.

⚠️ "Leaky gut syndrome" — as a diagnosis explaining fatigue, brain fog, joint pain, skin conditions, autoimmunity and food sensitivity — is a different claim.

Three problems with it:

1. ⚠️ The direction of causation is largely unestablished. Increased permeability is documented as a FEATURE of several diseases. Whether it causes them, results from them, or accompanies them is generally not known.

2. ⚠️ It is unfalsifiable as usually deployed. The symptom list is broad, the tests offered are unvalidated, and there is no defined threshold. Chapter 17 §17.1's fourth tell.

3. ⚠️ And the protocols sold to fix it — glutamine, collagen, bone broth, "gut healing" powders, elimination diets — have essentially no outcome evidence, while the elimination component carries §27.11's risks.

🔬 Verdict: 🟡 on intestinal permeability as a real, measurable phenomenon involved in defined diseases. ❌ on "leaky gut syndrome" as a diagnosis or as an explanation for non-specific symptoms.

⚠️ And Chapter 19 §19.11's lesson applies: dismissing the syndrome should not make you dismiss the physiology. Barrier function is real, it matters, and the best-supported thing you can do for it is §27.4 — which is butyrate, which is fibre.

27.12 Other gut conditions, briefly

⚠️ Because "gut health" collapses several distinct things.

Inflammatory bowel disease — Crohn's disease and ulcerative colitis. ⚠️ Diet does not cause IBD and diet does not cure it. But one dietary intervention is genuinely established: ✅ exclusive enteral nutrition — an exclusively liquid formula diet — is an effective induction therapy for paediatric Crohn's disease, ⚠️ with efficacy comparable to corticosteroids for inducing remission and without the steroid side effects. Most people find this surprising. The Crohn's Disease Exclusion Diet is an emerging alternative with growing evidence.

Coeliac disease — ⚠️ an autoimmune condition, not an intolerance; the treatment is strict lifelong gluten avoidance; and diagnosis requires testing BEFORE removing gluten. Chapter 28.

Diverticular disease — ⚠️ another stale-guidance case. Avoiding nuts, seeds and popcorn was standard advice for decades, on the theory that particles lodged in diverticula. When it was examined, no association was found, and the guidance was reversed. ⚠️ Many people are still following it. Fibre is associated with lower risk of developing diverticular disease.

Small intestinal bacterial overgrowth (SIBO) — ⚠️ a real entity that is over-diagnosed on the basis of breath tests with poor accuracy, and where the treatment is antibiotic rather than dietary. 🟡.

27.13 The two honest positions

⚠️ Part IV's device, brought back — because this is the one live controversy left in Part V.

⚠️ Both columns give identical practical advice: eat fermentable fibre and fermented food, don't buy the test.

Which is the seventh time in this book that the argument has been louder than the disagreement.

27.14 Who this chapter is about, and what to do

What to actually do, in order:

1. ⚠️ Eat more fermentable fibre. §27.4, Chapter 11. The whole system runs on it. 2. ⚠️ Add fermented food with live cultures, a serving or two daily. §27.9. Cheap, food, best randomized evidence here. 3. ⚠️ Don't buy the test. §27.3. 4. Use probiotics only for a named indication with a named strain, and expect the effect to stop when you stop. 5. ⚠️ If you have IBS, insist on the reintroduction phase. 6. And if you've been avoiding nuts, seeds and popcorn for diverticular disease — that advice was withdrawn.

7. ⚠️ Don't take antibiotics you don't need (§27.4b) — the most effective microbiome intervention available, and it isn't sold as one.

⚠️ A closing observation about that list. Six of the seven items cost nothing or save money, and the seventh is a short course of a named strain for a named indication.

There is no gut-health product on it. After a chapter about the most commercially active area in nutrition, the entire actionable output is: eat beans and oats, eat yoghurt or kefir, insist on reintroduction if you're restricting, and don't take antibiotics you don't need.

⚠️ That gap — between how much is sold here and how little is established — is the honest summary of the field as it currently stands. Which is not the same as saying it will stay that way.

🧾 What it costs

⚠️ The gap between price and evidence is wider here than anywhere in the book except Chapter 16.

⚠️ The two interventions with actual evidence — fermentable fibre and fermented food — are the two cheapest items on the page. Sauerkraut is cabbage and salt.

Theo's £179 would have bought roughly six years of home-made kefir.

Thirteenth chapter where the best-evidenced option is also the cheapest — ⚠️ and the only chapter where the expensive option actively recommended reducing the cheap one.

⚠️ How firmly I hold these

What we don't know

⚠️ Almost everything, and I'd rather say it plainly. Whether a "healthy microbiome" can be defined. Whether the associations with obesity, depression, autoimmunity and neurodegeneration are causal, consequential or confounded. Whether targeted manipulation will produce therapies for anything beyond C. difficile. ⚠️ And whether the fermented-food finding replicates at scale — which is the one I'd most want to know, because it's the one this chapter recommends acting on.

Spaced Review

1. (Chapter 11) Why does fibre keep reappearing in every chapter since Chapter 11?

⚠️ Because the microbiome runs on it. Fermentable fibre is the substrate for SCFA production — butyrate feeds colonocytes, and SCFAs support barrier function, immune regulation and satiety hormone release. ⚠️ Chapters 3, 7, 11, 22, 26 and 27 converge on the same recommendation from six different directions, which is what a real finding looks like approached from different fields.

2. (Chapter 19) §27.7 found probiotics delayed post-antibiotic recovery. Which earlier chapter has the same structure?

⚠️ Chapter 19 §19.5 — the omega-6 ratio. A clean mechanism, a confident recommendation, and a direct test that didn't go the expected way. "Take a probiotic after antibiotics" is the most common gut-health advice given, and the best mechanistic evidence suggests it may do the opposite. (⚠️ Held at moderate confidence — small mechanistic studies, not outcome trials.)

3. (Chapter 26) How does §26.11b's pattern apply to low-FODMAP?

⚠️ A motivated, symptomatic person is handed a restriction by someone who means well — and the restriction becomes the harm. Phase 1 works, so there's no symptomatic reason to progress, and people stay there for years: reduced fibre, reduced diversity, nutritional gaps, social restriction and disordered-eating risk. ⚠️ Low-FODMAP is a diagnostic process, not a diet.

Project Checkpoint: Your Fibre and Fermented Food Audit

Component twenty-seven. ⚠️ Do this before anyone sells you a probiotic.

Step 1 — Fibre, from Chapter 11's checkpoint.

Current intake: ____ g/day · Target: 25–30 g+ ⚠️ How much is FERMENTABLE? (beans, oats, barley, onions, garlic, leeks, bananas, cooled potato and rice, fruit and vegetables generally)

Step 2 — Fermented foods, live cultures only.

⚠️ Bread, beer, wine and vinegar pickles do not count — the microbes are gone or were never there.

Step 3 — The honest gap.

⚠️ §27.9's trial arm ate several servings a day. Most people start at close to zero. My realistic target: _ servings/day, of _____

Step 4 — ⚠️ The audit of what you're already taking or considering.

⚠️ If either middle column is "no," §27.6 says you're buying a species name.

Step 5 — And if you have IBS:

⚠️ Have I done first-line dietary advice? _ ⚠️ If I'm low-FODMAP: for how long? _ Have I done reintroduction? ____

⚠️ More than 6 weeks without reintroduction is the finding, and it needs a dietitian.

Next checkpoint (Chapter 28): your reaction log — what you actually react to, recorded properly, before you eliminate anything.

Chapter Summary

⚠️ The chapter's organizing distinction: what is ESTABLISHED versus what is INTERESTING. Both are real categories, they are not the same one, and almost every gut-health product depends on you not noticing.

⚠️ §27.4b's ranking inverts the marketing: antibiotics are the single most powerful modifier of your microbiome, long-term diet is second, and PROBIOTIC SUPPLEMENTS ARE LAST. ⚠️ And row 5 — medications including PPIs and metformin — means a substantial fraction of people who buy a test are measuring their prescription. Walt is on both.

⚠️ §27.3b: two labs disagree about one sample because of extraction method, 16S region, reference database, pipeline, handling and an unpublished comparison population. None of that is fraud — every step is a defensible choice. ⚠️ What you cannot do is take a number from one pipeline and read it as a fact about a person. The measurement is relative; the report presents it as absolute.

⚠️ §27.2's image: sequencing gives you a photograph of who was in the room, taken at the door, on one day. Not what they were doing, why, whether it mattered, or who was in the other rooms.

⚠️ §27.4 is the book's strongest structural argument: Chapters 3, 7, 11, 22, 26 and 27 converge on "eat more fermentable fibre from whole plant foods" from six independent directions. That's what a real finding looks like.

⚠️ §27.6's most important sentence: probiotic effects are STRAIN-SPECIFIC. Evidence for one strain transfers to nothing else — and this is almost never on the packaging.

⚠️ §27.7's inversion: most probiotics don't colonize; colonization resistance is personalized; and after antibiotics, probiotics DELAYED microbiome recovery relative to spontaneous recovery and to autologous transplant. The most common gut-health advice there is may do the opposite of what it intends.

⚠️ §27.11 is the most clinically useful section and the most often botched: low-FODMAP has THREE phases and people do the first one forever. It is a diagnostic process, not a diet — its purpose is to find your triggers so you can eat everything else.

The one thing to remember: ⚠️ Theo paid £179 to be told to eat fewer oats, apples and onions — which are the three foods most likely to have helped.

What's Next

Chapter 28 draws the distinction this chapter kept gesturing at: the difference between an immune reaction, an enzymatic one, and a reaction that isn't to the food at all.

IgE-mediated allergy and anaphylaxis. ⚠️ Coeliac disease — an autoimmune condition, and why you must be tested before removing gluten. Lactose intolerance, which is the global norm rather than the exception. Non-coeliac gluten sensitivity, where the honest answer is genuinely uncomfortable. ⚠️ Food intolerance tests — IgG panels, hair analysis, applied kinesiology — and what they actually detect.

And Alma's peanut allergy, seven years on — ⚠️ including what the last decade of immunotherapy has changed, and what it hasn't.