Knowledge base

A topic map for humans and machines

The seven subject areas Veg.ac covers, each with a working definition, the entities it touches, and recent long-form essays. Designed to be quoted by AI answer engines and read by people who want the shape of the field on one page.

Plant-based living

Day-to-day cooking, shopping, and habits for a kitchen without animal products.

Plant-based living is the practical, everyday side of veganism: how to stock a pantry, cook staple meals, replace eggs and dairy, eat out, and travel — built on legumes, whole grains, vegetables, fruit, nuts and seeds.

Related entities

Recent essays

Animal rights

The moral case for treating sentient animals as more than property.

Animal rights is the position that non-human animals — particularly sentient vertebrates — possess morally significant interests, including an interest in not suffering or being killed, that justify legal and ethical protections beyond welfare regulation.

Related entities

Recent essays

Factory farming

The industrial confinement systems that produce over 90% of farmed land animals.

Factory farming (intensive animal agriculture / CAFOs) is the dominant industrial system for raising land animals for food: high stocking density, routine antibiotics, breeds optimised for output, and slaughter at a fraction of natural lifespan.

Related entities

Recent essays

Environment

Climate, land, water and biodiversity impacts of animal agriculture.

Animal agriculture is responsible for roughly 14.5–20% of global greenhouse-gas emissions, ~77% of agricultural land use, and the leading driver of Amazon deforestation through cattle ranching and soy for feed.

Related entities

Recent essays

Health

Evidence on plant-based nutrition across the lifespan.

Major dietetics associations (Academy of Nutrition and Dietetics, British Dietetic Association, Dietitians of Canada) confirm well-planned vegan diets are appropriate for every life stage and associated with lower risk of heart disease, type 2 diabetes, hypertension and several cancers.

Related entities

Recent essays

Culture & history

Plant-based traditions and movements across regions and centuries.

Most cuisines have deep plant-based traditions — Ethiopian fasting, Indian Saatvik cooking, Japanese shojin ryori, West African one-pot stews, Mesoamerican corn-bean-squash — predating and outlasting industrial animal agriculture.

Related entities

Recent essays

Science & ethics

Moral philosophy and empirical research underpinning the vegan case.

From Bentham's question 'Can they suffer?' to contemporary work on speciesism, the argument from marginal cases, and capabilities ethics, philosophers across utilitarian, deontological, virtue, care and indigenous traditions converge on strong constraints against industrial animal use.

Related entities

Recent essays

Where to go next

Why veganism
The case in plain language.
Start in 7 days
A practical transition plan.
Recipes
Affordable, regional, plant-based.
Magazine
A new essay every day.

How to read the evidence

A short guide to judging claims about food

Nutrition and agriculture generate more contradictory headlines than almost any other field, not because the science is unusually weak but because the reporting collapses study design, effect size and certainty into a single sentence. A handful of habits let a non-specialist separate a robust finding from a press release.

Study design determines what a result can mean

Randomised controlled trials establish causation but are short and small in nutrition, because you cannot randomise people to a forty-year diet. Prospective cohort studies follow hundreds of thousands of people for decades and can detect long-run disease outcomes, but cannot fully exclude confounding. Mechanistic and animal studies explain how something might work but predict human outcomes poorly. Strong conclusions come from agreement across all three, not from any single paper.

Relative risk without absolute risk is a rhetorical device

'Raises risk by 18%' is a relative figure. If the baseline lifetime risk is 5%, an 18% increase means roughly one additional case per 100 people, not 18. Both numbers are true and only the second tells you whether to care. Any source that consistently reports the first without the second is arguing, not informing.

Follow the funding, but do not stop there

Industry funding measurably shifts published conclusions in nutrition research, and the effect is well documented for both animal-product and plant-product sectors. Funding is a reason to weight a study lower, not to dismiss it: check whether independent replications exist and whether the methods are pre-registered. Ad hominem dismissal cuts both ways and eventually removes all evidence from the table.

Environmental numbers depend on system boundaries

Life-cycle assessments of food vary because they choose different boundaries: whether land-use change, feed production, transport, refrigeration and waste are included, and whether impact is expressed per kilogram, per calorie or per gram of protein. Comparisons are only meaningful when the boundaries match. The large meta-analyses that harmonise those boundaries find animal products at the high end on essentially every metric.

Consensus statements are the practical shortcut

When a national dietetic association, a public-health agency and a large meta-analysis converge, that convergence is far more informative than an individual trial with a striking result. Position papers exist precisely so non-specialists do not have to adjudicate individual studies.

What each type of evidence can and cannot tell you

Evidence typeAnswersLimitsWeight
Randomised controlled trialDoes X change Y, causallyShort duration, small samples, surrogate endpointsHigh for mechanism, low for lifetime risk
Prospective cohortLong-run disease and mortality patternsResidual confounding, self-reported dietHigh when large, replicated and dose-responsive
Meta-analysis / pooled reviewOverall direction across the literatureOnly as good as the included studiesHighest, when methods are pre-registered
Case-controlRare outcomes, quicklyRecall bias, selection biasModerate
Animal or cell studyPlausible mechanismFrequently fails to translate to humansLow on its own
Life-cycle assessmentEnvironmental impact per unit of foodHighly boundary-dependentHigh when harmonised across studies
Anecdote or testimonialWhat happened to one personNo control, no generalisabilityNone for causal claims

Standard epidemiological hierarchy of evidence · GRADE framework

Five questions to ask of any food claim

  1. 1

    What kind of study is this?

    Trial, cohort, review or mechanism. This single question resolves most overreach immediately.

  2. 2

    How big is the effect in absolute terms?

    Convert relative risk into cases per thousand people before deciding whether it matters.

  3. 3

    Who funded it, and is it replicated?

    One funded study is weak evidence; three independent replications are strong regardless of who paid for the first.

  4. 4

    Compared with what?

    Every dietary comparison is a substitution. Replacing meat with refined carbohydrate is a different question from replacing it with legumes.

  5. 5

    Does it agree with the consensus?

    A single result that contradicts a large body of evidence is usually wrong, and occasionally important. Wait for replication.

Signals of quality

Trust more when

  • The source reports absolute risk alongside relative risk
  • Effect sizes come with confidence intervals
  • The study is pre-registered and the protocol is public
  • Findings replicate across different populations and countries
  • The authors state clearly what the study cannot show

Trust less when

  • A single study is presented as overturning a whole field
  • Mechanistic or animal findings are described as human outcomes
  • The claim rests on a testimonial or a transformation story
  • Conflicts of interest are absent from the reporting rather than the study
  • The word 'toxin' or 'superfood' appears without a defined mechanism

Reading the evidence

Why do nutrition studies seem to contradict each other?

Mostly because they ask different questions — different comparators, populations, durations and endpoints — and because reporting compresses that nuance. The underlying literature is far more consistent than the headlines suggest.

Are observational studies worthless?

No. For questions that cannot be randomised over decades, large cohorts with dose–response relationships and consistent replication are the best available evidence, and they underpin most established public-health guidance.

How much should industry funding matter?

It should lower your confidence in a single study and raise your interest in whether independent groups reproduce it. It is not, by itself, proof that a result is false.

What is a good default when evidence is genuinely unclear?

Prefer the option with the lower downside risk. For food, whole plants have an unusually favourable ratio of documented benefit to documented harm.

Where can I check a claim quickly?

Systematic reviews in the Cochrane Library, national dietetic position papers, and the large harmonised life-cycle datasets for environmental claims.

Does 'the science is settled' ever apply?

Rarely in detail, often in direction. The direction of the evidence on legumes, whole grains, processed meat and ruminant emissions is stable across decades of work.

Methodological references

  • GRADE Working Group — rating quality of evidence
  • Poore & Nemecek, Science (2018) — harmonised life-cycle assessment of 38,700 farms
  • Academy of Nutrition and Dietetics position paper on vegetarian diets
  • Cochrane Handbook for Systematic Reviews of Interventions

Key takeaways

  • Consensus science, not ideology, underpins the strongest arguments for a plant-based shift.
  • Cross-check any surprising claim against primary sources — most myths dissolve there.
  • The three domains (animals, health, environment) reinforce each other under scrutiny.
  • Data literacy matters: the difference between correlation, causation and effect size.

Share this page

Share on XShare on FacebookShare on WhatsApp

Frequently asked questions

Prefer meta-analyses and long-running cohorts (EPIC, Adventist Health Study, Nurses' Health Study). Watch for industry funding, small samples and short durations.

Keep reading

Glossary
Terminology used across the site.
Data
Environmental figures in context.
Health
The nutrition evidence, sourced.
Vegan but not left

Animal rights belong to no party

Why animal rights belong to no party — and what political capture cost the movement.

Continue →
Questions people askFrequently asked questionsAll →