Biohacking — self-directed experimentation to optimize biology — has moved from fringe subculture to mainstream conversation. A significant subset involves research compounds: peptides, nootropics, longevity molecules, and metabolic modulators. Understanding the evidence landscape requires distinguishing rigorous science from preliminary data and anecdote. This overview specifically addresses the science of biohacking, and related research is indexed on PubMed.
Levels of Evidence
Not all evidence is equal. A useful hierarchy for evaluating research compounds: systematic reviews and meta-analyses (highest confidence), randomized controlled trials (gold standard for human causation), prospective cohort studies (observational human data), controlled animal studies (mechanistic value; limited human translation), in vitro cell culture (hypothesis-generating), and case reports or anecdote (signal-generating only).
Many popular research compounds have evidence concentrated in the lower tiers. That does not make the research meaningless — it reflects where the compound is in the research pipeline.
Compounds with Stronger Evidence Bases
Some compounds studied in biohacking communities have accumulated RCT data: semaglutide and tirzepatide (GLP-1 receptor agonists), metformin for metabolic health markers, rapamycin (mTOR inhibition) in aging models, and NMN/NR (NAD+ precursors) in small human trials.
Compounds with Primarily Preclinical Evidence
Most peptides popular in biohacking — BPC-157, TB-500, CJC-1295, Ipamorelin, Epithalon, Selank — have predominantly animal model data. Animal models can reveal genuine biology, but human evidence is absent or very preliminary, and risks in humans are not well characterized.
Explore evidence-graded compound profiles at the compound database and peptide research hub.
Cognitive Biases That Distort Self-Experimentation
Beyond the evidence-tier hierarchy, self-experimentation (N-of-1, meaning a sample size of one) carries specific, well-documented cognitive traps worth naming directly:
- Placebo response — expectation of benefit measurably changes subjective outcomes (energy, mood, pain, perceived recovery) even with an inert substance, and this effect is well-documented to be strong for exactly the kind of subjective outcomes (sleep quality, “feeling recovered,” mental clarity) that biohacking self-reports commonly track.
- Regression to the mean — people tend to start a new protocol when they feel unusually bad (a rough patch, a flare-up, a bad week), and symptoms often improve on their own afterward regardless of intervention, which can be misattributed to whatever was started at the low point.
- Confirmation bias in self-tracking — when you expect a compound to help sleep, you may notice and remember the good nights more than the average ones, especially without a structured, blinded tracking protocol.
- Survivorship bias in community reporting — people who had a strong positive experience are more likely to post about it than people who felt nothing or had a negative experience, which skews the visible community narrative toward positive outcomes regardless of the compound’s true average effect.
- Publication bias in the underlying literature itself — studies with positive, statistically significant findings are more likely to be submitted and accepted for publication than null-result studies, which can make a compound’s published literature look more consistently positive than its true underlying evidence base.
A Practical Self-Tracking Discipline
For anyone conducting genuine self-directed research (as opposed to casual use), a few disciplines meaningfully improve the quality of self-generated data: track a defined outcome measure consistently before starting any intervention to establish a real baseline (not a retrospective guess at “how things were before”), track for a defined period rather than stopping as soon as a desired change appears, avoid changing multiple variables simultaneously (a common practice that makes it impossible to attribute any observed change to a specific compound), and be honest that N-of-1 self-tracking — however carefully done — cannot control for placebo response or regression to the mean the way a properly blinded, controlled study can. This doesn’t make self-tracking worthless; it makes it hypothesis-generating for yourself, not proof for others.
Frequently Asked Questions
If I feel a strong effect from a compound, doesn’t that prove it works for me?
It demonstrates that you experienced a change while using it, but without a controlled comparison it can’t distinguish a real pharmacological effect from placebo response, regression to the mean, or an unrelated lifestyle change happening at the same time.
Why do online community reports skew so positive for many research compounds?
Survivorship bias in what people choose to post, combined with the placebo and confirmation-bias effects described above, tends to make visible community sentiment more uniformly positive than the underlying evidence base actually supports.
Does publication bias mean published studies can’t be trusted?
Not that they can’t be trusted, but that the aggregate published literature for a given compound may skew more positive than its true average effect, which is one reason systematic reviews (which explicitly attempt to account for this) sit at the top of the evidence hierarchy described above.
Educational only. Not a recommendation to use any research compound or engage in self-experimentation.
