hawkeyefan
Legend
When in the real world you're standing in front of a steep bank or cliff that you might or might not be able to climb, how closely do you know your odds of being able to climb it?
Pretty good, actually. I think that people are generally better at this than is claimed in these kinds of discussions. Especially highly trained people.
I used to play soccer a lot. I could tell you with reliably well what was within my capabilities and what was in doubt and what was impossible. I could assess my skill compared to that of other players. I knew my weaknesses and my strengths as a player.
I also used to ski quite a bit. I knew what I could handle on a mountain. I recognized when my skill level increased and I became more comfortable tackling tougher slopes.
I think that generally speaking, trained people have a solid understanding of their skill and the challenge posed by an obstacle. Enough to portray that with rudimentary math and a randomization method, for sure.
I'd bet it would come down to estimates along the lines of "piece of cake", "shouldn't be a problem", "could be tricky", "yeah, that's a bit risky", "I'm only trying that if it's life or death", and "not a hope". Usually the first and last of those - "piece of cake" and "not a hope" - are pretty obvious, but the rest aren't so clear. Further, those perceptions might not always be accurate - "this looked easy, but now I'm stuck", or "I thought this would be tough but now I'm up here there's way more handholds than I could see from the bottom." - and giving a hard number denies these misperceptions.
Sure. If only there was some tool we could use to randomize the results to allow for this. It would be ideal if the tool was small and could be held in hand and cast onto a table so that the results could be easily seen by all participants.
If only!
That's how I want the players-as-characters thinking (and as a player, that's how I want to be thinking). The DM, be it me or someone else, can worry about the actual mechanics behind the scenes.
Hard numbers ruin that for two reasons: one, they break immersion; but (and IMO more importantly) two, they don't allow me to make a mistake in perception or interpretation. If for whatever reason I've talked myself into thinking this cliff will be easy to climb and the DM then throws out that its DC is 19, that meta-information is going to make me rethink in a hurry. Flip side - if I've talked myself into thinking I can't climb it but then the DM says the DC is only 10, up I go.
Some numbers are unavoidable. Others (such as task DCs) have a choice as to whether they can be replaced with description, and my stance is that it's better to do it that way and leave some uncertainty (as would be the case in reality) before even getting to the roll.
Numbers don't break immersion nearly as much as not being able to accurately understand the scenario. When that happens, I'm immediately reminded that I'm removed from the situation.
A mistake in perception can be represented by the dice. This is why your whole idea of "certainty" just doesn't work for me... it's not certain because we still have to roll the die, and none of us can predict what the roll will be.
Metagaming means - and always has - considering and (ab)using elements outside what the character can perceive.
What is it that's shared that's not based on what the character perceives?