Post 2 of 3 on uncertainty, probability and confidence
They're certainly
vague ranges and sometimes quite arbitrary, I'll give you that. But as highlighted in the post above, my point all along has been that
the alternative is even more arbitrary to the point of both introducing major biases and being patently wrong in many cases: Refusing to guestimate and justify and refine degrees (or percentages) of confidence in a proposition doesn't mean you're avoiding doing so, it means that the degrees of confidence you are implicitly holding are either a 1 or a 0, 'believe' or 'don't believe.' Consequently rather than trying to
weigh or critically evaluate all available evidence and scenarios under consideration, one inclined towards one side or the other whether 'believe' or 'don't believe' burdens themselves with a bias towards
stacking up favourable evidence on their preferred side and downplaying the evidence on the other... and usually without any clear criteria of what is needed to "cross the line" and swap one's position between one side or the other.
Conversely attempts to guestimate and justify and refine degrees of confidence needn't be entirely arbitrary, by any stretch of the imagination. If one supposes (hopefully for justifiable reasons) that the scenario A is twice as likely as all the main scenarios constituting not-A, then mathematically one should have roughly 66% confidence in A, or a bit less in deference to unknown/other scenarios. Or in the case I outlined in that post, where A and not-A are perfectly
equivalent variables due to A being nonspecific and irreducible, we should necessarily have equal 50/50 confidence in either one. That's known as the
principle of indifference incidentally, and we all use it more or less intuitively; if I gave you a 12-sided die you wouldn't need to roll it thousands of times and tally the result to work out the probability of rolling a 4. It can be generalized as
for any set of N equivalent, mutually exclusive and collectively exhaustive variables the probability of one outcome (or for more complex real-world examples, our Bayesian level of confidence in one scenario) is 1/N.
If you've got a murder victim and four main suspects, absent any more specific evidence a rational person would view each suspect as ~20-30% "likely" to be the killer. It's not precise or perfect of course, but it's obviously not "pulling numbers out of the air." These guestimates are somewhat vague or arbitrary because our knowledge is somewhat vague or arbitrary - more on that in my post below.
As far as I can tell the most common reason that "no onus, no evidence, no belief" atheists are uncomfortable with that style of thinking is that they share with many religious folk a preference for absolutes ("no evidence!") and discomfort with nuance, uncertainty or lack of clarity, preferring the clear dividing lines of belief or nonbelief.