It's a way of getting to a binary result, a "yes/no", based on a mix of induction (sufficient weight of authority demonstrating the principle) and deduction from it.

What it isn't is an assessment of statistical likelihood in a population. It always starts in the individual facts of one case.

April 09, 2026 - 23:00 UTC

1
0
1
That "is/isn't" switch means what what we don't do very well is things like sensitivity v. specificity, or p values, or that the existence of an individual outlier in a given case does not invalidate the statistical analysis of a population *because* they are an outlier. We don't think statistically
1
0
2
In short, I don't think we understand - certainly don't use - statistical reasoning about populations or how scientific papers work, and they don't get our focus on an individual case.

It means that in cases where it turns on population analysis, it ends up getting messy, because we don't do that.
1
0
1