"correcting" weapon vs armor type from chainmail

Delta recently wrote an excellent post analyzing the origin of the Greyhawk/AD&D weapon vs armor type table and concluding (rightly, in my view) that it is fundamentally flawed. I just wanted to offer a few follow-up analyses of ways that one could try to "correct" the modifiers in there.

Putting the tables below the break.

To summarize Delta's work (though you should really read it for yourself, it's good), the weapon vs armor modifiers come from simply subtracting the to-hit numbers on the Chainmail weapon vs armor table:

From a flat modifier of 8, which is very close to the average value for this whole table. The problem with this approach is that the different armor types should have different base values, since more armor in general offers greater protection. That's enough for Delta to dismiss the Greyhawk/AD&D tables. However it made me curious what the weapon vs armor table would look like if one used an average based on the column rather than the whole table. That yields the following tables:

I don't know if anyone should use these tables, but here they are. This is how much better or worse "than average" each weapon is in Chainmail.

Or is it?

Of course, numerical modifiers on a 2d6 table do not translate directly to percentile differences. To analyze things further, I converted the Chainmail weapon vs armor table to the equivalent decimal number you'd need to roll-over on a d20 to match the probability:

Then, we can directly compare compare these target numbers to the average for each armor type and to the d20 rolls needed to hit that armor class in actual D&D; I chose a THAC0 of 19, somewhere between the 18 of a 3rd or 4th level fighter and the 20 of a 1st or 2nd level fighter, rather than the value of 21 which would probably be used for Chainmail's "normal men," since those target numbers are closest to the average values for each armor type. The results follow:

Again, I don't really think anyone should play with these weapon/armor modifiers? They don't strike me as particularly realistic (maces are not particularly well-known to be less effective against unarmored foes) or balanced (what do you have against spears, Gary), and they make the basic combat mechanic of the game significantly more complicated.