Thanks for that detailed analysis, perrinmiller. That's exactly the kind of feedback I was looking for. I find a few flaws with it though, mostly not in the math, but in the assumptions. First of all, you assume that 1st-level is probably where the impact will be felt the most. I happen to disagree. I've been playing in a RotR campaign for several months now with a GM who runs things 99% from the book. 1st-level all the way through 8th-level we've encountered some battles where no one was hitting on a reliable basis. When we were, we were hitting hard, but it wasn't very productive sometimes for 2-3 rounds in a row. We had similar issues in 3.5 when running the Shackled City AP, Age of Worms AP, and Savage Tide AP (two of those with me as GM). It usually isn't until levels 13+ when to-hit starts becoming a non-issue. With my slight modification of the Armor as DR, even a non-damaging hit is converted to 1 nonlethal, which is still progress in the majority of situations. Our experiences are clearly different though, so your point remains partially valid.
You also made a flaw in your calculations. The average damage from a shortbow hitting a goblin is actually:
(1 nonlethal + 1 nonlethal + 1 lethal + 2 lethal + 3 lethal + 4 lethal) / 6 = 2 (given the assumption that a monster does not heal). Also, the factor of critical hits becomes a more substantial contribution to average damage as AC goes down (although the math in this situation is different because crits follow different rules so that might not actually be the case; I'd have to modify the spreadsheet I made for calculating average damage that I use to determine when it is optimal to use things like Power Attack and smites). The math isn't as simple as (avg. die roll - DR) for monsters capable of taking nonlethal damage.
I am wondering why we need a different rule variant for magical attacks. If we need the GM to provide meta-information on Saving Throws to self resolve magical spells anyway, then why not just allow players to roll the saving throw for the GM in that case?
I have no objection to that. Do players object to me rolling their saves during monster turns in the interest of speeding up game play? (Exceptions could be made for saving throws versus death or debilitating effects like feeblemind). This does shift the dynamic of Hero Points more towards offensive use, though you could still use them for the +4 or reroll when your turn comes up to reverse the outcome. (Imagine the movie where The Hero is battling The Villain in the final fight and The Villain strikes what appears to be a telling blow. The Hero collapses to the ground, appearing out for the count; The Villain has won and begins to cackle maniacally as lightning crackles above. But then The Hero lifts his arm and plants his fist in the mud (it's raining you see), draws himself up, and The Villain turns around just in time to be cut down with a mighty blow from The Hero's sword. This could be a dramatic example of The Hero using a Hero Point to reverse an apparent outcome.)
Kinem switched to wizard, so we have plenty of arcane support but we are now slightly weaker on melee strength.
Good point.
That all looks good to me, airwalkrr!
The only suggestion I have is regarding knowledge checks. Since all of the knowledge is assumed to be there in the characters' heads, searching for it isn't really like finding the right folder on a computer. So in my games, I don't require the players to specify which knowledge skill they're using or roll separately for whichever skill might apply - they just make a single d20 roll, and I apply the appropriate bonus from their sheet (or from their stat block, if they've got it detailed there). This seems to speed things along a little.
Meh, that's more work on my part, and GMs have enough work. I don't see why expecting a character trained in a specific Knowledge skill wouldn't be able to recognize an undead type or a magical beast type and recall the basic characteristics of type at the very least. I get the sense I don't fully grasp your point though.