Why Personal Self-Experiments Rarely Settle Anything

Self-experimentation is the defining practice of biohacking, and it produces genuine information about an individual. The obstacles to interpreting that information are structural rather than a matter of effort.

Everything changes at once

A person adopting a new practice rarely changes only that. Sleep timing, meal composition, activity and stress shift alongside it, often without being noticed.

Trials in research settings handle this by randomising participants, so that unmeasured factors are distributed evenly between groups. A single person cannot be randomised against themselves in the same way.

The consequence is that any observed improvement has multiple candidate explanations. The intervention is one of them, but it is not privileged over the others by the data.

People start when things are worst

Interventions are typically adopted at a low point, when symptoms or measurements are unusually poor. That timing has statistical consequences.

Measurements that fluctuate around an average tend to move back towards that average after an extreme reading. This is regression to the mean, and it produces improvement without any cause.

The effect is strongest for the most variable measures, which are often the ones people track most eagerly. Mood, energy and sleep quality all fluctuate substantially day to day.

Expectation shapes what is recorded

Someone who has invested effort and money into a practice is not a neutral observer of its effects. Expectation influences both physiological response and how experience is recorded.

Blinding exists in research precisely to break this link, and it is difficult to arrange for oneself. Most interventions announce themselves through taste, sensation or visible equipment.

Subjective outcomes are most affected, but objective ones are not immune, since expectation influences behaviour on the days a measurement is taken.

Sequence effects distort comparison

Comparing a period before an intervention against a period after it assumes nothing else differed between those periods. Seasons, workload and illness rarely cooperate.

Alternating between the intervention and its absence repeatedly, in an order decided in advance, addresses part of this. Repetition separates a consistent effect from a one-off coincidence.

Carryover complicates even that design. If an effect persists after the practice stops, the comparison periods are contaminated, and the measured difference shrinks.

What self-experiments can establish

These constraints do not make personal experimentation worthless. They limit it to a particular type of conclusion, which is about the individual rather than about the intervention.

A practice consistently followed by a measurable change over many alternating cycles is reasonable evidence for that person, without establishing why or generalising to anyone else.

Where a self-experiment involves symptoms, prescribed medication or anything that could mask a developing problem, the appropriate step is discussion with a clinician rather than more data collection.