Research methods · Putting it to work

Is it strong enough to act?

The whole track so far has trained you to ask is this claim true? — to walk a headline back to its hedged original, to spot the survey masquerading as a trial, to distrust the dramatic anecdote. That skill leaves you somewhere uncomfortable: you now see that the evidence for almost everything is messier than it's sold as. Most of what you'd consider for a cut, a build, or daily life sits in a fog of small studies and competing meta-analyses. If you wait for proof before acting, you'll never act. So the messy truth has to become a clean decision somehow — and that needs a second question the first one can't answer.

Two axes, not one

The mistake is to sort everything onto a single line from "unproven" to "proven" and act only on the proven end. That throws away cheap, harmless wins and — worse — it lets a confident claim pull you into something costly or irreversible before the evidence has earned it. Decisions live on two axes at once. One is how strong the evidence is. The other is what acting costs you if you're wrong — money, yes, but mostly risk and reversibility. A bad week of a cheap powder you stop taking is nothing. A liver injury, a botched cut that costs you muscle, or a year on the wrong protocol is not something you simply undo.

Read the grid as four standing instructions. The teal corner is the only one where the answer is unconditional.

Asymmetry is what "is it true?" leaves out

Truth is a probability; asymmetry is the shape of being wrong. Two interventions can have the same 30% chance of helping and call for opposite decisions, because one costs you a few dollars and a shrug while the other costs you a body part you can't grow back. Three questions pin the asymmetry down. Cost: what do I spend, in money and effort? Risk: what's the realistic worst case, and how likely? Reversibility: if it's not working — or quietly harming me — can I notice and stop before damage is done? Cheap-safe-reversible is the magic combination that lets you act on thin evidence, because the bad outcome is small and undoable. Lose any one of those and the bar for evidence climbs.

If you try, run an honest n-of-1

"Maybe try" is not "swallow it and vibe." A self-experiment — an n-of-1 — is the lower-left quadrant done properly, and two failure modes wreck the lazy version. Placebo: expecting a thing to work makes you feel it working, especially on soft measures like energy and mood. Regression to the mean: you start most things on a bad week, and bad weeks are followed by average ones no matter what you took — so anything looks like it helped. The fix is to decide the rules before you start.

• Pre-specify the measure. Pick something you'd struggle to fool yourself on — a logged lift, morning weight trend, grip strength — not "I feel sharper."
• Pre-specify what "it worked" means. Write the number down first. "Bench moves up 5kg over 6 weeks," not a verdict you reverse-engineer later.
• Run long enough, ideally on and off. One good week proves nothing. Stopping and seeing the gain vanish is the strongest signal a solo experiment can give you.

Work one, then finish one

Worked: creatine monohydrate vs an unproven "longevity" compound. Creatine has decades of trials and meta-analyses behind it for strength and lean mass, costs a few dollars a month, and has a clean safety record in healthy people — strong evidence, low cost, low risk. Lower-right corner: just do it; it doesn't need an experiment. The longevity compound rests on mouse data and a couple of small human studies, costs $180 a month, and has unknown long-term effects on a body you can't swap out — weak evidence, high cost, real downside. Upper-left corner: skip. Notice the evidence didn't have to be settled to decide; the asymmetry did the work.

Your turn: a supplement is cheap, plainly safe, biologically plausible, but genuinely unproven — say a few small trials hint it helps recovery, nothing conclusive. Where does it land, and what do you do? (Lower-left — weak evidence, low cost and risk. Reasonable to try, but as a pre-specified n-of-1: decide upfront what you'll measure, e.g. logged session volume or soreness on a fixed scale, set the number that counts as "worked," run it on and off for several weeks, and stay alert to placebo and regression to the mean.)

Why this matters

This is the payoff of the whole track — the move that turns "the evidence is messy" into a decision you can actually make about your own body. Across a real cut → build → maintain plan the questions arrive constantly, and the grid sorts them fast. The GLP-1 drug for the aggressive cut is upper-right: it carries real risk, but it's prescribed, monitored by a clinician, and backed by strong trial evidence — worth it, with eyes open, not something to improvise. A protein target and progressive overload for the build are lower-right: strong evidence, low risk, just do it. A trending fat-burner stack some podcaster swears by is usually upper-left — stimulants and unknowns for a thin, hyped claim — so skip. And a cheap, harmless recovery aid is lower-left: fine to self-test, on your terms, with the number written down first. The same four boxes carry you from the cut into maintenance without ever needing the evidence to be clean. This is educational, not medical advice; anything involving a prescription drug is a conversation with your doctor.