Science loves control because reality is badly behaved.
A clinical trial can standardize dose, screening, timing, environment and measurement. That is precisely why clinical trials are useful. They reduce the number of things that can explain an outcome.
Most psychedelic use, unfortunately for tidy diagrams, does not occur inside a clinical trial.
That gap is the premise of a new project at the University of California, San Francisco. ARPA-H’s UCSF Canopy Project, which began September 21, 2026 with up to $1.4 million in funding, is designed to collect what the agency calls “ecologically valid” data about psychedelic effects on the brain and body in more complex real-world environments.
The principal investigator is Robin Carhart-Harris, one of the best-known researchers in modern psychedelic neuroscience.
Why controlled data and real-world data answer different questions
Imagine you want to know whether a car can brake safely.
A test track is useful because you can control speed, surface and obstacles. But eventually you also need to know what happens in rain, traffic, potholes, bad decisions and the magnificent improvisation of other drivers.
Psychedelic research has a similar problem.
Controlled studies can tell us about a defined compound, dose, population and protocol. They are essential for testing medical claims. But outside studies, people use substances of uncertain dose and purity, in social settings, ceremonies, retreats, bedrooms, festivals and therapeutic contexts that may not resemble formal trials at all.
Those environments introduce confounding variables by the truckload.
They also contain the reality policymakers are being asked to govern.
“Real world” does not automatically mean “more true”
There is a fashionable tendency to treat naturalistic data as if removing laboratory walls reveals reality in its pure form.
It does not.
Real-world data are messy because the world is full of self-selection, memory errors, co-use of substances, unequal access, changing environments, incomplete records and people who do not behave like study protocols.
A controlled trial may sacrifice generalizability for causal clarity. An observational dataset may gain breadth while losing the ability to say exactly what caused what.
The interesting science comes from using both rather than forcing one to impersonate the other.
What the Canopy Project says it wants to do
According to ARPA-H, the project aims to capture brain and body data at scale in complex environments while retaining prospective scientific rigor. The stated goal is to generate insight that can help mitigate risk and inform policy.
That is a different ambition from asking whether a drug treats a specific disorder.
It moves the question toward public health: what actually happens when people use psychedelics outside the protected geometry of a trial?
The answer could become especially important as laws, clinical programs and public interest evolve faster than long-term datasets.
The privacy problem is not a footnote
The more intimate the data, the more useful and dangerous the dataset can become.
Psychedelic experiences can involve mental-health information, physiological measurements, location, behavior, subjective reports and potentially wearable or digital traces. Collecting rich naturalistic data therefore creates a second scientific problem: how to study people without turning unusual states of consciousness into a surveillance product.
The public description of the project is high level, so details of implementation and privacy protections will matter as the work develops.
That is not a reason to reject real-world research. It is a reason to pay attention to study design with the same intensity normally reserved for the headline result.
Better evidence may make the story less simple
There is a possibility the psychedelic field should welcome: better real-world data may make both enthusiasts and critics less comfortable.
Some benefits seen in controlled settings may prove harder to reproduce amid uncontrolled conditions. Some feared harms may prove uncommon. Others may cluster in specific substances, populations or environments. Context may explain more than broad labels like “psychedelic” allow.
That is what useful evidence often does. It destroys a simple argument and replaces it with a map.
Psychedelic culture has had plenty of manifestos. It can survive a few more maps.