The thing with that is that it's actually a useful generalization to make in a lot of scenarios.

If you know nothing about the distinction between two possible outcomes, treating them as equally likely is a helpful tool to continue with the back of the envelope guess. Knowing this path needs 5 coin tosses to go right and this one needs 10 is helpful to approximate which is better.

Your example is obviously outside the realm where you have zero information, so uniform distribution is no longer the reasonable default. But the idea is from a reasonable technique, taken to extremes by someone who doesn't fully get it.