Despite hitting the hay a bit early last night (and tossing up a reserve post to maintain my crunch time post quota) I've made a substantial bit of progress. I've added all the revisions, rewrites and new material since my original draft many moons ago. All I have left for now is the equipment section and a few clarifying sections for the basic prototype. Now that I've spoken of my success let me share one of my more recent failures.
A while back I was reading a post over on Delta's D&D Hotspot that was referencing a posting over at myArmory forums about a gentleman who decided to test some authentic or equivalent medieval weapons against cheaply made chainmail as well as a handcrafted set of the highest quality. It's an excellent posting and experiment be sure to check it out when you have the time. The test showed how different European weapons, as well as a katana thrown in to highlight a few differences, would perform against the most common of medieval armors, chainmail and jack (layered textile most closely related to brigadine.) The tests provided some excellent insights and I thought I might veer more towards simulation side of game taxonomy  for my weapons system. Mind you, this was before people declared with great fortitude their staunch distaste for weapons vs armor class in Jeff Rient's poll. Furthermore, I had already taken a stab at this before and the results were more trouble than they were worth. Still I was sure this time would be different.
I may not have mentioned this before but I use abstract damage for my games where all weapons do a d6 and two-handers do 2d6. I already have a method in place for differentiating individual weapons but I also wanted a way to differentiate them via weapon class or perhaps damage type, whether it be cutting, thrusting or impact. I was also experimenting with an armor system at the time and I decided to place that burden on armor, rather than adding another layer to weapons. First we'll start with the armor system.
Armor comes in three divisions, light, medium, and heavy with a rating of 1,2, & 3 respectively. This rating determined how many six-sided dice you would roll. If a single die roll was a 5 or 6, damage was halved; if two or more dice were 5 or 6 then damage was nullified entirely. It was simple and compact and I liked it, little did I know I was merely adding a twist to the standard all or nothing battleship armor of standard D&D so it was a bit futile to begin with. The system became increasingly complex when we started adding in weapon types, and which armor was more effective against them based on the experiment performed over at myArmory. Here's what I had originally.
Jack: +1 vs Bows, -1 vs Cutting
Mail: +1 vs Cutting, -1 vs Thrusting
Plate: +1 vs Cutting, +1 vs Axes, +1 vs Arrows, -1 vs Impact
There was also a set for leather and hide but those were based off of speculation and not empiricism. The +1 and -1 would be added to each d6 you roll, making you more or less likely to negate damage in such a way. Right off the bat we have pluses and minuses which is a bit too fiddly for my tastes so those were dropped completely. We also have +1 vs Axe but -1 vs Impact (Axes deliver force through an impact rather than a cutting motion) which is curious to the average onlooker. The idea being that on the knightly poleaxe, the hammer was better suited for platemail as the axe could slide across the plate. In fact the little teeth showing on the hammer in the picture of the poleaxe are there because the teeth make it more likely to catch the plate thus delivering the force with minimal loss. Makes plenty of sense but it was still too fiddly. Here is the revised form:
Jack: +1 vs Ranged.
Mail: +1 vs Cutting
Plate: +1 vs Cutting and Thrusting
Plate picked up Thrusting to simulate Impact being the preferred form for overcoming it. Also, most thrusting weapons would feasibly skitter off the surface if they didn't find a joint, except ones specifically designed for dealing with plate (the rondell dagger and perhaps the estoc.) It was a bit more streamlined, but as we discovered in playtesting, it didn't provide enough lasting benefit while being worn.
If we're looking at mail the probability of halving the damage is (1/6 + 1/6) or 1/3, the probability of both being 6 and nullifying the damage is (1/6 * 1/6) is a 1/36 chance. So a reasonable amount of the time it will halve damage (33%), and on rare occasions (2.78%) it will stop all the damage. If it were against a cutting weapon the percentages increase to approximately 55% and 11% respectively but that's not very indicative of the tests performed where chainmail more or less completely stopped a cutting a weapon nearly all of the time.
Then there is the player's perception which views armor as a bit of a crapshoot as the protection it provides and the amount of time it actually works does not seem to be worth the investment. I tried another method where instead of having a +1 to your roll, if your favorable against that damage type you were treated as having an automatic success. It worked a fair bit better but it still fell short of player expectations. So it was scrapped and later recycled for shields which fulfill a different purpose in the player's mindset. We'll see in time how well that fares.
 Regrettably I cannot recall the husband and wife duo that first published the gaming taxonomy triangle that I first encountered in my upper level coursework.