It feels too dismissive.

While I do agree, that:

- there is a chance of something other causing the instant relief AND matching the moment of cannabis use

- health policy for population, should be made based on studies of population

At the same time, we must accept also some limitations of medical trials.

Models that interpret gathered results always includes a random part. Why? World is quite deterministic, why the randomness? Because one can’t make all possible measurements (money, sample size, time), one must choose the most promising practical setup.

Imagine hypothetical situation:

- there are 30 genotypes in population

- drug is highly effective only for 1 of those

- study doesn’t make genetic testing (also, it’s a parallel group study)

Such setup inflates required sample size to get statistically significant results. And even if significance is found, it will say that effectiveness is only 1/30, so not that good of a drug.

(30 is not the limit, think of a case with 300 types or 3000 types)

Human body is amazingly complex. It is not a solved problem.

If OP experienced instant relief of pain after smoking cannabis, it would be a logical action for OP to try it second time if pain reappears. (Given that cost/risk of such personal experiment is relatively low)

> there is a chance of something other causing the instant relief AND matching the moment of cannabis use

You forgot about selection bias here

The point is not it's impossible to deduce. The point was that without randomization even simple passage of time gets equivalent support from that experience as cannabis. People make that mistake all the time with common cold, cause they drink some special tea on day 7, get the final relief, and think that was the tea.