Self-Experimentation Risk

An uncontrolled self-experiment is a single observer, unblinded, with an expectation about the result, measuring a subjective outcome against a moving baseline. Nearly every mechanism that clinical trial design exists to defeat is present at once: placebo response, regression to the mean, the natural fluctuation of the condition, seasonal and behavioural change, and the tendency to start something new at a low point when improvement was likely anyway.

The physical side is separate and specific. A compound bought outside a regulated supply chain has unverified identity, purity and endotoxin content, no assurance that the vial contains what the label says or contains only that, and no manufacturing oversight behind the sterility of the powder. Nothing collects what goes wrong afterwards, so events are neither counted nor communicated to anyone who could act on them.

The instructive contrast is that self-experiments are not worthless in principle. The 1984 gastritis self-experiment that helped establish Helicobacter pylori as a cause of ulcers produced usable evidence because the endpoint was objective and independently verifiable by biopsy and culture, not because the experimenter felt different. A personal trial with a pre-specified objective measure, blinded product and randomised order can genuinely inform one person about one thing.

The transfer error is where the harm compounds. A result in one unblinded person cannot support a claim about a population, yet forum aggregates of such reports are read as though they were a trial. Those aggregates are also filtered by survivorship, since people who had a bad experience or simply stopped rarely post again, leaving a record that is systematically more positive than the underlying experience.