Scientific Knowledge Has a Provenance

A person explaining physiology on social media did not discover human physiology by opening a camera.

A reel about insulin, cortisol, inflammation, aspartame, muscle protein synthesis, mitochondrial function or cardiovascular risk depends on a body of knowledge that existed before the reel and was produced elsewhere.

That observation is not an appeal to academic status.

It is a question of provenance.

Where did the claim come from?

Which experiments, measurements, clinical observations, mathematical models, reviews or regulatory assessments made the claim possible?

Who established the terminology?

Who measured the quantities?

Who determined the uncertainty?

Who found the exceptions?

Who showed that one mechanism mattered more than another?

These questions become especially important when a communicator presents scientific institutions as irrelevant, corrupt, detached from reality or inferior to personal experience while continuing to use knowledge created by those same research systems.

The contradiction is not that academia can be criticised.

It should be criticised.

The contradiction appears when institutional authority is rejected but institutional knowledge is silently retained as personal authority.

Scientific knowledge is distributed

Academia is not the sole producer of scientific knowledge.

Universities are one part of a much larger system that includes:

• hospitals
• government laboratories
• public health agencies
• regulatory bodies
• private research laboratories
• pharmaceutical and biotechnology companies
• engineering firms
• professional societies
• standards organisations
• independent research institutes

A serious criticism of "academia" therefore does not remove dependence on organised scientific research.

Modern knowledge is distributed across specialised communities.

A toxicologist relies on analytical chemistry.

A clinician relies on laboratory medicine.

A nutrition researcher relies on physiology, epidemiology and biostatistics.

A statistician analysing a clinical trial relies on the validity of the clinical measurements.

A public communicator sits even further downstream.

John Hardwig described this dependence in his work on epistemic dependence. Much of modern knowledge is too specialised for one person to verify from first principles. Scientists themselves rely on the competence and testimony of other experts because no individual can personally reproduce every experiment, instrument calibration, mathematical derivation and data collection process on which contemporary science rests.

Scientific knowledge is therefore not a collection of isolated personal discoveries.

It is a network.

The communicator usually enters near the end of the chain

Consider a simplified chain:

[ R longrightarrow A longrightarrow S longrightarrow C, ]

where

[ R ]

represents raw observations,

[ A ]

represents analysis,

[ S ]

represents synthesis across studies,

and

[ C ]

represents the public claim.

For example:

[ ext{laboratory measurements}

ightarrow ext{study estimate}

ightarrow ext{systematic review}

ightarrow ext{public explanation}. ]

A communicator normally operates at the final stage.

That is not a weakness.

Division of labour is what makes modern science possible.

The problem begins when the final stage is presented as though the preceding stages did not matter.

A confident explanation can make a scientific claim sound like direct personal knowledge even when nearly every substantive component came from other people's work.

Provenance is different from authority

Suppose a communicator says:

Substance X is metabolised into compounds A and B.

The truth of that claim does not depend on whether the communicator has a PhD.

It depends on whether the claim is supported by biochemical evidence.

A non-academic can state it correctly.

A professor can state it incorrectly.

The important question is whether the claim can be traced to evidence.

Scientific provenance therefore avoids two bad arguments at once.

The first is:

[ ext{no academic credential} Rightarrow ext{claim is false}. ]

That is invalid.

The second is:

[ ext{confident communicator with visible results} Rightarrow ext{claim is scientifically established}. ]

That is also invalid.

The speaker's identity and the claim's provenance are separate variables.

A source is useful only if it supports the claim actually being made

Citation is not enough.

Suppose a study establishes

[ C_1: ext{A biomarker changed under a particular intervention in mice.} ]

A public post concludes

[ C_2: ext{The intervention prevents disease in humans.} ]

The citation is real.

The inferential bridge is not.

The transformation from (C_1) to (C_2) has introduced several additional assumptions:

[ ext{mouse}

ightarrow ext{human}, ]

[ ext{biomarker}

ightarrow ext{clinical outcome}, ]

[ ext{experimental dose}

ightarrow ext{ordinary exposure}, ]

and possibly

[ ext{association}

ightarrow ext{causal effect}. ]

The scientific problem is therefore not simply whether a source exists.

It is whether the source supports the strength and scope of the public claim.

Every transformation can lose information

Let the original evidence support a set of scientifically defensible claims

[ mathcal A(E). ]

A public claim (C) is adequately supported only if

[ Cinmathcal A(E). ]

Communication requires compression.

A paper may contain forty pages of methods, qualifications, uncertainty and alternative interpretations. A public explanation may have three paragraphs or ninety seconds.

Some information must disappear.

The question is which information.

Removing technical details that do not change interpretation is communication.

Removing the conditions that determine whether the conclusion is true is distortion.

The most important omissions are often invisible

A claim can survive grammatically while changing scientifically.

Consider:

In this study, among adults with condition X, intervention Y reduced biomarker Z by an average of 8% over twelve weeks.

Now compress it to:

Y reduces Z.

The shorter sentence has removed:

• the population
• the duration
• the outcome definition
• the magnitude
• the uncertainty
• the comparison group
• the study design

Compress it again:

Y improves health.

Now the biomarker has become a health outcome.

The final sentence may sound clearer.

It is scientifically less specific.

A good communicator simplifies language without silently strengthening the conclusion.

Scientific vocabulary is inherited vocabulary

Terms such as:

• insulin resistance
• oxidative stress
• systemic inflammation
• endocrine disruption
• autophagy
• mitochondrial dysfunction
• gut permeability
• neuroplasticity

did not originate as marketing concepts.

They come from scientific literatures with operational definitions, measurement problems, competing models and domain-specific limitations.

Using the terminology imports that history.

A communicator cannot reasonably use the prestige of the concept while discarding the methodological constraints that give the concept meaning.

If "inflammation" is invoked, the next question is which inflammatory process and how it was measured.

If "toxicity" is invoked, dose and exposure matter.

If "insulin resistance" is invoked, the method used to assess it matters.

Scientific vocabulary is not a licence to skip scientific specification.

Personal results do not create the underlying scientific theory

A coach can discover useful practice through experience.

A trainer may notice that a particular programming strategy works well with clients.

A clinician may recognise a recurring pattern before formal research catches up.

Practical knowledge matters.

But practical success does not independently establish biochemical or physiological explanations.

Suppose a training programme produces good results.

That observation can support claims about feasibility and observed outcomes.

It does not establish that the programme worked because of a specific molecular pathway unless the pathway was itself investigated.

The explanation still depends on external scientific knowledge.

This is why "I see the results" cannot replace the question:

How do you know why those results occurred?

Outcome and mechanism are different scientific claims.

Criticising academia is not anti-scientific

Universities and journals have real problems.

Publication bias exists.

Replication failures exist.

Statistical incentives can reward novelty over reliability.

Peer review can miss important errors.

Prestigious institutions can defend weak ideas.

Researchers can have conflicts of interest.

Academic status can create social authority that exceeds evidential authority.

None of those criticisms is controversial in serious methodological literature.

The Open Science movement exists largely because researchers criticised weaknesses inside scientific practice.

Ioannidis' widely discussed 2005 paper on false findings was not an attack from outside science. It was scientific criticism of scientific practice.

The existence of institutional failure therefore does not justify abandoning the institutions' outputs indiscriminately.

It justifies better methods for evaluating them.

Science contains mechanisms for criticising itself

One reason organised science is valuable is not that it never fails.

It is that it contains procedures through which failure can become visible.

These include:

• replication
• reanalysis
• methodological criticism
• peer commentary
• correction
• retraction
• systematic review
• competing laboratories
• improved measurement
• new experiments

The procedures do not operate perfectly.

They still create an important asymmetry between a claim embedded in a research literature and a claim supported only by the confidence of its speaker.

A public communicator can be wrong without any formal mechanism forcing the error to be revisited.

A research claim at least enters a system where other people can inspect the methods and attempt to contradict it.

Peer review is a filter, not a truth machine

Defending provenance does not require romanticising peer review.

A peer-reviewed paper can be poor.

A preprint can be excellent.

A textbook can contain outdated material.

A systematic review can inherit the biases of its included studies.

A regulatory assessment can be revised.

The value of provenance is not that one institutional label guarantees truth.

The value is that the reader can reconstruct the chain.

A paper provides methods.

A review provides inclusion criteria.

A guideline describes how evidence was weighed.

A public communicator should provide enough information for the audience to find those objects.

Provenance creates auditability.

Expertise matters because interpretation is not mechanical

If everyone can read a paper, why should expertise matter?

Because scientific interpretation is not just extracting the conclusion sentence.

A specialist knows which measurements are trusted, which approximations are controversial, which effects are large enough to matter, which methods routinely fail and which apparently new result conflicts with a much larger literature.

This is tacit knowledge as much as formal knowledge.

Collins and Evans distinguish forms of expertise partly to explain why participation in a technical community changes what a person can reliably understand and judge.

That does not make specialists infallible.

It explains why reading five abstracts is not equivalent to years spent working inside a field.

Cross-disciplinary confidence should fall, not remain constant

A person can be highly competent in one domain and poorly informed in another.

This is normal.

Scientific expertise is narrow because the knowledge base is large.

The warning sign is not crossing disciplinary boundaries.

Scientists do that constantly.

The warning sign is confidence that remains unchanged while the subject changes.

A fitness coach discussing exercise programming may be operating inside deep practical expertise.

The same person discussing endocrine pathology, toxicology, psychiatric medication and cancer epidemiology has crossed several professional and scientific boundaries.

The communication standard should become more conservative as distance from expertise increases.

More sourcing is needed.

More qualification is needed.

More willingness to defer to specialist evidence is needed.

Popularity measures distribution, not validity

An audience of one million people is evidence of reach.

It may be evidence of communication skill.

It can be evidence of commercial demand.

It is not an estimator of scientific correctness.

Let

[ N ]

be audience size and

[ T ]

be truth of a scientific claim.

There is no general reason to expect

[ P(Tmid N ext{ large}) ]

to be close to one.

Social platforms optimise engagement, not epistemic calibration.

Confidence, simplicity, emotional relevance and visual clarity can all improve distribution.

Scientific uncertainty can do the opposite.

A communication environment can therefore reward properties that are only weakly related to scientific reliability.

Commercial success creates another source of asymmetry

Suppose a communicator sells a programme.

If the programme succeeds commercially, the communicator gains:

• revenue
• testimonials
• audience growth
• social proof
• more opportunities to repeat the explanation

If the scientific explanation is wrong but the programme still produces some satisfied clients, there may be little commercial pressure to correct the mechanism.

The market validates the product proposition.

It does not automatically validate the scientific proposition.

This distinction is important because financial success can feel like empirical confirmation.

It is confirmation of demand for the service.

That is not the same estimand.

A source chain can be audited

For a scientific claim in public communication, a useful provenance chain is:

[ C leftarrow S leftarrow P leftarrow D, ]

where

[ C ]

is the public claim,

[ S ]

is the synthesis or interpretation,

[ P ]

is primary research,

and

[ D ]

is the underlying data or experimental observation.

Not every claim needs to be traced all the way to raw data.

That would make communication impossible.

But the chain should exist.

For well-established knowledge, a high-quality review, textbook or professional assessment may be the appropriate source.

For a new or controversial claim, direct primary evidence becomes more important.

The level of sourcing should match the uncertainty and novelty of the claim.

Secondary sources are not automatically weaker

A common misunderstanding is that primary papers are always superior sources.

For many public claims, a systematic review or carefully constructed guideline is more informative than one primary study because it evaluates a body of evidence.

The question is not:

Is this the original paper?

The question is:

Is this the appropriate evidence object for the claim?

A molecular mechanism may require primary experimental work.

A claim about average clinical efficacy may be better supported by a meta-analysis.

A claim about recommended intake may require a regulatory or expert assessment that integrates several evidence streams.

Provenance is about fit between source and claim, not proximity to the laboratory bench.

Cherry-picking breaks provenance even when every citation is genuine

Suppose ten studies exist.

Eight estimate effects near zero.

Two estimate large positive effects.

A communicator cites the two positive studies accurately.

Every citation is real.

The resulting picture of the literature is still misleading.

Source accuracy and source selection are separate issues.

A scientifically responsible communicator should therefore ask not only:

Does this source support my statement?

but also:

Why did I choose this source rather than the rest of the relevant evidence?

This is one reason systematic evidence synthesis matters.

It constrains selective citation.

The strongest public claim should not exceed the strongest defensible source

Suppose the best available evidence supports:

X may modestly reduce outcome Y in population P.

A communicator states:

X fixes Y.

No amount of confidence in delivery repairs the mismatch.

A simple discipline is:

[ ext{Strength(public claim)} le ext{Strength(evidence)}. ]

This is not a literal numerical inequality.

It is a constraint on scientific communication.

The more uncertain the evidence, the more conditional the language should become.

The more restricted the population, the narrower the generalisation.

The more indirect the outcome, the more cautious the practical conclusion.

Uncertainty is part of the source

When a paper reports

[ hat heta=0.30 ]

with a wide uncertainty interval, the interval is not optional metadata.

It is part of what the study learned.

Communicating only the point estimate changes the information object.

Likewise, if a systematic review reports high heterogeneity, that heterogeneity belongs to the conclusion.

If evidence quality is low, the limitation belongs to the claim.

A communicator who retains only the direction of the result is not merely simplifying.

They are changing the evidence.

Good science communication exposes where the communicator added judgement

Suppose the literature contains mixed evidence.

A communicator concludes that one interpretation is more persuasive.

That can be legitimate.

The important distinction is between:

[ ext{the evidence reports} ]

and

[ ext{I infer}. ]

Public communication becomes more trustworthy when those layers are visible.

For example:

The review found small average effects with substantial heterogeneity. I think the most likely explanation is X because of Y and Z.

That sentence exposes the judgement.

A weaker form silently converts the interpretation into fact.

"Do your own research" has a limit

Independent reading is valuable.

It can improve scientific literacy and help people challenge weak claims.

But modern science cannot be rebuilt personally from first principles.

To independently verify an endocrine claim, one may need expertise in:

• physiology
• assay methodology
• statistics
• pharmacokinetics
• clinical epidemiology

To verify a toxicology claim, one may need:

• analytical chemistry
• exposure assessment
• dose response modelling
• pathology
• regulatory toxicology

At some point, everyone depends on specialised communities.

The relevant skill is not eliminating dependence.

It is managing dependence intelligently.

Epistemic independence is not the absence of trust

A scientifically independent thinker does not trust nobody.

They know what kinds of trust are warranted.

They prefer claims that are:

• traceable
• testable
• open to criticism
• replicated where possible
• supported by appropriate expertise
• revised when evidence changes

Independence means being able to inspect why a claim deserves trust.

It does not mean pretending to have personally generated every piece of knowledge involved.

Institutions should earn trust through transparency

None of this implies that universities or scientific organisations deserve unconditional trust.

Trust should be connected to process.

An institution deserves more epistemic trust when it makes methods visible, discloses conflicts, publishes uncertainty, allows criticism, shares data when possible and corrects mistakes.

Institutional prestige without transparency is weak evidence.

The same standard should apply to public communicators.

The difference is that scientific institutions are often expected to document provenance as part of normal practice.

Public platforms rarely require it.

A communicator can reject institutional status while preserving provenance

There is no requirement to defer rhetorically to academia.

A communicator can say:

Universities sometimes reward bad incentives.

That can be true.

They can say:

Peer review misses errors.

Also true.

They can criticise publication practices, funding systems, academic gatekeeping or weak statistical standards.

None of that requires pretending that scientific knowledge appeared independently of organised research.

A coherent position is:

The institutions are imperfect, so I want the evidence to be more transparent and auditable.

An incoherent position is:

The institutions are worthless, but I will continue presenting their accumulated findings as evidence of my own scientific authority.

The question to ask is simple

When someone makes a scientific claim in a reel, podcast, post or sales page, the first question does not need to be:

What are their credentials?

A better first question is:

Where did this knowledge come from?

Then follow the chain.

What study?

What review?

What textbook?

What consensus report?

What population?

What dose?

What outcome?

What uncertainty?

What alternative evidence?

If the claim is established science, the provenance should be easy to identify.

If it is a personal interpretation, that should be clear.

If it is a new claim, the evidential burden should increase.

If no source can be identified, the audience is not being asked to evaluate science.

It is being asked to trust a speaker.

Scientific authority should remain detachable from the personality

The strongest scientific communication has a useful property.

The communicator can disappear and the argument still stands.

The claim can be checked without the audience needing to admire the person's physique, income, confidence, credentials, followers or lifestyle.

The evidence exists independently.

That is the advantage of provenance.

It moves authority away from personality and back toward an auditable chain of observations, analysis and criticism.

A public communicator does not weaken their authority by showing where their knowledge came from.

They make the authority more legitimate.

Scientific knowledge has a history.

Good communication does not erase it.

References

Collins, H., & Evans, R. (2007). Rethinking Expertise. University of Chicago Press.

Hardwig, J. (1985). Epistemic dependence. The Journal of Philosophy, 82(7), 335–349. https://doi.org/10.2307/2026523

Ioannidis, J. P. A. (2005). Why most published research findings are false. PLoS Medicine, 2(8), e124. https://doi.org/10.1371/journal.pmed.0020124

Longino, H. E. (1990). Science as Social Knowledge: Values and Objectivity in Scientific Inquiry. Princeton University Press.

National Academies of Sciences, Engineering, and Medicine. (2017). Communicating Science Effectively: A Research Agenda. Washington, DC: The National Academies Press. https://doi.org/10.17226/23674

Oreskes, N. (2019). Why Trust Science? Princeton University Press.