Finding the Real Experts

Every field has two populations of expert. There are the people who do the work — run the trials, publish the papers, disclose the numbers — and there are the people who explain the work to everyone else. The second group is vastly larger, vastly more visible, and optimised for something other than being right.

This is not a complaint about popularisers. It's a search problem. When you go looking for who to learn from, the ranking you get back is sorted by reach, and reach correlates with production quality and confidence far more than it correlates with accuracy. The most-viewed source in almost any domain is, structurally, not the most reliable one.

So I built a filter. It started as a paragraph I was retyping into every research conversation — some version of "give me the people who work from primary literature, not the biggest channels" — and it kept working, so I formalised it.

---

#The two axes people constantly merge

The single most useful move is separating two questions that get answered as one:

How good is the evidence? A meta-analysis and a mouse study are different objects.

Who is telling me, and what do they want? A company reporting on its own product and an independent researcher reporting on the same product are different objects.

These are orthogonal. A company-sourced claim about a well-designed RCT and a researcher-sourced claim about animal data are different problems, and collapsing them into one "is this trustworthy" judgement loses the information you needed.

Almost all bad health, fitness and business advice is not fabricated. It's real research, stripped of its effect size, its confidence interval and its limitations, by someone whose income depends on it sounding certain.

#The tier that actually matters

The filter has six tiers, but the interesting one is the third: the researcher-practitioner. Someone who works directly from primary literature, publishes their own analysis, and engages with evidence rather than citing it decoratively.

In hypertrophy that's Chris Beardsley — he reasons at the level of motor unit recruitment and crossbridge mechanics, publishes his own preprints, and will tell you when a mechanism he'd like to be true is under-evidenced. In sprint mechanics it's Jean-Benoît Morin, who publishes his methods and his spreadsheets openly. In aging biology it's Matt Kaeberlein, which is the rare case where the most credible person in a subfield is also the one communicating publicly.

None of them are the biggest name in their space. That's not a coincidence — it's close to a diagnostic.

#Six ways to fail

The disqualification pass matters more than the ranking. A candidate is dropped if they:

Cite without engaging. A study named on screen, its effect size never given, its limitations never mentioned. This is the most common failure and the hardest to spot, because it looks exactly like rigour.

Sell the certainty. The confidence outruns the literature, and there's a course or a supplement line at the end that the confidence was manufacturing demand for.

Drift outside their field. Expertise doesn't transfer. A brilliant molecular biologist opining on nutrition policy is a surface-tier source in that moment.

Never publicly change their mind. A spotless record in a live field is evidence of not paying attention.

Claim operator status without numbers. A physique, a lifestyle, a client list, or a good framework is not a disclosed, dated, verifiable record. In business content specifically, almost nobody clears this bar, and the handful who do — public revenue dashboards, decade-long disclosed figures — are worth more than the rest combined.

Only agree with their own audience. Audience capture is indistinguishable from expertise when you're inside it.

#The correction that keeps the whole thing honest

There's a failure mode built into any filter like this, and it took me a while to see it: the filter can start selecting for obscurity.

If you build a system that penalises popularity, it will eventually hand you a contrarian answer because contrarian answers feel like the product of rigorous filtering. That's the same error as credulity, wearing better clothes.

So the calibration is explicit, and it overrides everything else: well-known is not the same as weak. If the strongest evidence in a domain points at something unglamorous that everyone has heard of — sleep, sunscreen, protein, index funds, walking, flossing — the correct output is to say so plainly and explain the mechanism that makes it work.

Depth means digging past the first answer. It does not mean rejecting the first answer because it was first.

#Naming the gaps

The last piece is the one I didn't expect to need. Sometimes the honest output is that no trustworthy communicator exists.

Skincare is the standing example. I've looked repeatedly, and there is no single source in consumer skincare that reliably avoids blending real citations with product marketing. The correct answer is not to nominate the least-bad option — it's to say the gap exists and point at the primary literature instead.

A filter that always returns a name is a filter that will eventually return a bad one.

---

#The prompt

Here it is in full. Paste it, name a domain, and it returns the tier structure rather than the top ten channels. It's also installed locally as a Claude skill, so it applies automatically whenever a source-credibility question comes up rather than needing to be invoked.

Paste this, then name a domain. It returns Beardsley-tier sources instead of popularisers, in any field, without retyping the calibration.

---

#THE PROMPT

Find me the best sources to learn [DOMAIN] from.

Apply the source-tier discipline below. Do not skip the disqualification pass, and do not pad the list to look thorough — a five-name answer where every name survives scrutiny beats a twenty-name answer.

#What I'm asking for

Rank by whose claims can actually be trusted and for what, not by popularity, follower count, or production quality. I want the people the popularisers are quietly reading.

#The tiers

#The calibration test

The canonical example: for hypertrophy, the correct answer is Chris Beardsley (tier 3 — reasons at the level of motor unit recruitment and crossbridge mechanics, publishes preprints, states when a mechanism is unproven), plus the publishing academics running the actual trials (Milo Wolf, Pak Androulakis-Korakakis, Brad Schoenfeld, Eric Helms — tier 2), plus Greg Nuckols for statistical literacy.

The wrong answer is a high-production "science-based" YouTube channel. If your output looks like a list of the largest channels in the space, you have failed and should restart.

#Disqualification pass — run this on every candidate before including them

Drop anyone who:

1. Cites without engaging. Names a study on screen; never gives the effect size, the sample, the limitations, or what would falsify it.
2. Sells the certainty. The confidence outruns the literature, and there is a course, a coaching package, or a supplement line at the end that the confidence was manufacturing demand for.
3. Has drifted outside their field. Expertise does not transfer. A brilliant molecular biologist opining on nutrition policy is a tier-5 source in that moment.
4. Never publicly changes their mind. A track record with no reversals in a live field is evidence of not paying attention, not of being right.
5. Claims operator status without numbers. A physique, a lifestyle, a client list, or a good framework is not a disclosed, dated, verifiable record.
6. Only ever agrees with the consensus of their own audience. Audience capture reads exactly like expertise from the inside.

#What to return

For each source, give me:

• Name, tier, and the one-line reason for that tier — specifically what makes them tier 3 rather than tier 5.
• What they're actually good for, and what they cannot support.
• Where their best material is, and honestly whether it's paywalled. Say so rather than pretending you read it.
• Their known blind spot or bias — everyone has one; if you can't name it you haven't read enough of them.
• Who credibly disagrees with them, and on what. A field where all the good sources agree is a field you're not seeing clearly.

Then separately:

• The disqualified list — 2–4 prominent names in this domain that fail the pass, with the specific failure. Naming these is as useful as naming the good ones, because they're what I'd otherwise find first.
• Gaps. If no trustworthy communicator exists for a sub-area, say so and point me at the primary literature instead. Naming a gap beats nominating a mediocre substitute. (Skincare is the standing example — there genuinely isn't one.)
• The single best entry point. One paper, book, or article to start with. Prefer the primary source over any summary of it.

#Two calibrations that override the instinct to be clever

"Well-known" is not the same as "weak." If the strongest evidence in this domain points at something unglamorous that everyone has heard of, say so plainly and explain the mechanism. Sleep, sunscreen, protein, index funds, and walking are all correct answers that sound boring. Skepticism belongs on evidence quality, never on popularity.

Depth means rigour, not contrarianism. Do not manufacture a counterintuitive answer to seem sophisticated. Reflexive iconoclasm is the same failure as credulity, wearing better clothes.

---

#Already-run domains

Cached so they don't need re-running. Re-run any of these annually — sources get captured, drift, or start selling something.