Let's not. In medicine, N=1 is no joke: the reason being that, when (i) there are so many unknowns, (ii) there is so little experimental data, and (iii) the con

Let's not. In medicine, N=1 is no joke: the reason being that, when (i) there are so many unknowns, (ii) there is so little experimental data, and (iii) the consequences are so potentially severe, from a decision-theoretical point of view it makes no sense to ignore data at all, no matter how biased. You would be surprised how many of the 'black-box' warnings in common pharmaceuticals come from on N=1 individual reports. Hell, papers are written on individual medical cases... Now you might even argue that hearsay is not 'data', but in fact it is: for instance, there is quite a bit of NLP literature on trying to identify unreported side effects of medications from 'informal' channels like social media. Even if it's a second or third or fourth report, it is a report nonetheless, and should be taken seriously. So let's not be gratuitiously negative [1] about people's negative reports about marihuana, no matter how biased you might think they might be. This is one of the most dramatic public health changes since the Prohibition we're talking about, so what is at stake is the health of millions of people for possibly many generations --so even very small effects otherwise considered too rare to matter might have an outsize impact on large segments of society. [1] http://blog.ycombinator.com/new-hacker-news-guideline http://blog.ycombinator.com/new-hacker-news-guideline