Someone asked me if I had taken RuneQuest's rules for damage-absorption armor and adapted them for D&D. In terms of my game design history, I owe a lot to RuneQuest. Still, there are two aspects of the RuneQuest system that I do not use in my RPG designs, and damage-absorption armor is one of them. The other is roll-under skill checks.
Roll-Under Skill Checks
Percentile skill checks entered the RPG universe with thieves' skills in Greyhawk (1975). In this system, they were no skill checks opposed to each other. No one made one skill check against another, such as "Spot" against "Hide." Listen checks were made on d6 and Move Silently checks on percentile, with no implication that they should relate to each other.
RuneQuest took the percentile skill idea and ran with it. Everyone could try (almost) every skill with a base percentile chance (depending on the particular skill), a modifier (depending on ability modifiers), and bonuses for talent, training, or experience. The problem RQ had was handling skill vs. skill contests, with a secondary problem of environmental factors (difficulty). The system worked great as long as a simple dice check against your skill level meant success. Once modifiers for the environment and skill checks against an opponent came into play, it got a little weird. What happens when a master silent mover (95%) meets a master listener (95%)? Is it the same as an amateur silent mover against an amateur listener (25% each)? If your Search skill is 50%, do you have the same chance to find a clue that's been dropped casually as one that's been hidden? What if you're looking in the dark?
RuneQuest 3rd edition added a host of modifiers to address environmental or circumstantial factors. Pendragon added the tie-breaker of "how high did you get without going over" your skill, and BESM used the mathematically identical but semantically reversed tie-breaker of "how much did you beat the roll by." Both of these systems are mathematically very close to "die + bonus" systems, but less straightforward.
After playing a lot of RuneQuest and Call of Cthulhu (same system), plus some Pendragon, I concluded that "die + bonus" was the way to go. Mark ReinHagen and I released a "die + bonus" system in Ars Magica, about the same time that Mike Pondsmith designed a die + bonus system for Cyber-Punk and Stephen Michael Sechi did the same for Talislanta. Once I joined the 3rd edition team for D&D, I championed the die + bonus system that is now the core mechanic of the game.
HeroQuest uses a "roll under" mechanic, akin to Pendragon, with the same breakpoint problems that such systems are heir to. When I run HeroQuest, I simply convert it into a mathematically compatible and smoother die + bonus system.
Armor absorption means that armor reduces incoming damage instead of turning hits into outright misses. Armor absorption makes characters more brittle. As the amount of armor you have rises relative to your hit points, wounding blows become less common. More hits become practical misses and deal no damage. The GM needs to ratchet up damage to make encounters challenging. But hits that are capable of overcoming armor are also likely to deal lots of damage relative to your hit points. The range of damage that hurts you but doesn't take you out of the fight becomes narrower, and hits become closer to binary (nothing/knock-out).
Tangentially, you see the same brittleness with the neg-10 rule in D&D. As your hit points get higher, and damage gets higher, the range between "OK" and "dead" becomes practically smaller. That's why I use the rule that you don't die until you reach negative "10 + level."
Character brittleness is especially pronounced in RuneQuest, where hit points are almost static, hits are applied to individual hit locations (each with a fraction of total hit points), and criticals and impales let you unpredictably get past armor.
The D&D system isn't realistic. Imagine using a shield to block a giant's club. The club would break the shield and the character's arm. But the system is very playable. When hits have a small chance to take a character out of the fight and a small chance to weaken the defender, that's a more unstable, unpredictable system than when hits have a 100% chance to weaken the defender and a very small chance (if any) of taking the defender out of the fight. Predictability is bad for realism and good for game play.
HeroQuest doesn't have persistent damage, so a fight's only relevant if there's a chance that the PCs will be defeated (high unpredictability). Robin dealt with the high unpredictability by instilling an expectation that defeated opponents are usually not killed. Thus the high unpredictability of the combat (you don't know who'll win) doesn't lead to TPKs (total party kills).
A hypothetical armor-absorption system that reduced damage by a percentage would avoid the problem of making characters brittle. The only practical strike against it is the amount of math involved. Chivalry & Sorcery has a system where armor's damage absorption was rolled for each blow, the way one normally rolls for damage. That has the benefit of allowing smaller amounts of damage to remain relevant and reduces the pressure on the GM to increase opponents' damage capacity.