Not sure how to start given you say the game wasn't designed for high level play. I guess? I mean I'm trying to fix that a bit with some homebrew. Leveling up is fun after a big milestone, and if the campaign becomes ridiculous past level 10-15, what is the point of the system? So, aside from the other reason I gave (easier NPC/monster creation), lets say the goals are: (a) higher level play that is more rewarding when you level (i.e. getting +3 forever feels better than a lifetime of +1 and perhaps +0.5 bonuses), (b) making it so a level 15 fighter would seriously think twice about fighting a level 20 fighter because those extra 10 ranks make for another +30 OB (at +3 each regardless of rank number) and a +120 vs +90 is significant in battle as opposed to +80 vs +85 in the existing system, and (c) a slightly slower progression so characters aren't able to easily overcome difficult actions.
We come from different schools of RPGing, I guess. I have no problem with a level 15 fighter trying to take on a level 20 fighter he doesn't know - because he doesn't know them. And if they know them, then I'll be sure to give them an in-character estimation of their would-be opponent's skill.
I am also of the school that level is not a major factor all the time. Level class is, to some extend - say, 'low' levels (1-4), 'medium' level (5-10) and 'high' level (11-20). The difference mostly comes from intelligent use of abilities and non-combat spells (investigation and utility). I tend to think of level 20 as a glass ceiling: most people never go above that, because you need more than adventuring and XPs to reach beyond that threshold - you need to understand how the world works, what the power players are, and decide what role you want to play in that game. And then, you have to survive level 50+ people suddenly paying attention.
So, level 15 is not very different from level 18. And I second that. But level 25 is *very* different from level 20, and it's not a question of skill ranks and available spells. So I don't need those to mark the difference.
You addressed the third point with some examples that were interesting but slightly confusing to me. Are you saying that physical feats are not absolute? Climbing a slick wall in the rain isn't the same difficulty regardless of the person trying it? Are you saying the difficulty should be like a D&D 5e DC level and just auto scale with higher level characters? If you are not saying this, then my point about +3 flat being slightly more interesting and better for lower level play still holds. They will notice over time their ability to do more difficult things they have failed at many times before. If you are indeed saying all things should scale, then I guess I disagree. There should indeed be a different between party members and what they are capable of, and if everything scaled to the most skilled person then everyone else has probably wasted their ranks.
No, difficulties are not relative. They are absolute, but I don't use the scale provided in the RM rules because the names are confusing ('easy' ? 'hard' ? Easy for who and hard for who ?), and the modifiers do not come with a proper rationale. I treat all manoeuvers as opposed manoeuvers, with a simple rule of thumb: if someone is opposed by another person with a similar skill bonus, the chance of success is 50%. So, instead of saying that a manoeuver is "hard" or "easy", I try to come up with the skill bonus that would create such an opposition, or, if it is a natural obstacle (such as a cliff to climb), what kind of skill bonus would lead to a 50-50 chances of success. And then, I can rate the difficulty appropriately (the difficulty modifier is equal to 50 minus this skill bonus).
Then, the only thing I need to do is to rate the typical NPCs: what skill bonus does an 'apprentice' have ? What about a 'professional', or a 'regional champion' ? And with that kind of rating table, I have all the difficulty modifiers I need: a manoeuver is not 'hard', it's one that would give a professional (+70 skill bonus) a run for their money (50-50 odds), so the difficulty is -20.
I use such a table. I also have a small table that gives me difficulty modifiers for effort, and another table that gives me modifiers for conditions and props. These are very generic and more than enough for almost all situations... which I round up to 'all situations'.
So, basically, having a linear ROI for skill ranks developed instead of the RM tradition of having diminishing return is not something I use, because I don't consider level as being a major factor in determining the actual skill of a person. Rather, level determines one's understanding of the 'true world' and one's place in this world. Skills are skills, and since their progression is level-independent (I don't use level bonuses), I have no reason to use a linear modifier. I would come back to a linear progression, but then, I would probably move the diminishing ROI on the skill development cost (rank acquisition cost becoming increasingly more expensive the greater the number of skill ranks you already have in the skill). Though, to be fair, I would probably use a similar system for spell list acquisition, because right now, RM uses diminishing return for skill rank bonus but there is no diminishing return for spell learning - actually, there is a very steep decrease in difficulty after level 20 since the next spell (level 25) actually costs five times as much as the previous one, but it is linear again after that (and you can see where my interpretation of level 20 as a glass ceiling is coming from - in RM1, it would likely have been level 15).
I also tend to run campaigns with a well-defined experience ceiling. Characters start at a given level (say, 6) and I tell the players what level they can expect to reach by the end of the campaign. I had a level 4-6 campaign, a level 6-10, a level 12-19, a level 20-20 (five years of play, not a single level gain - that's the glass ceiling), and the current one is level 9-12. This makes things tighter.
