Mike and I have begun generating the content that is going to go into the Swords of Kos RPG, using my Swords of Infinity rules engine, and during this process I've found some gaps that needed addressing. One of the most grievous gaps in the rules became apparent as I was creating stats for a simple rat, one of the creatures that will end up in the RPG. The issue was with my size rules, which modified a monster's ability scores by a positive or negative amount depending on its size category. The bonuses and penalties can be found in the following table. 

These modifiers seemed relatively straight-forward to me, balanced, and, for the size spread given in the table, these rules worked well! But I said that these rules were problematic, and the rat helped me identify why. According to the rules as written, a rat would qualify as a size -20 (not a typo), and following the pattern laid out in my table, that would give the poor rat a -200 modifier to Strength and Vitality, and a +200 modifier to Dexterity and Awareness (in a percentile based system this puts us way outside the regular bounds). So this hyper-aware, hyper-agile rat would be the weakest and sickliest creature in the world. My rules were modeling a reality that did not jive with the feeling of the rest of the game. Two rats fighting each other would never be able to hurt each other, and cats would never be able to catch or sneak up on the damn things.

The adjustment I made to the rules were simple, and based on the conclusion that size only really makes a difference to creatures of a different size -- my rules needed to be relative. I ended up creating a new table, one that lists modifiers based on the size category difference between two creatures that are interacting with each other. It's a simple table, the difference in size is multiplied by 10, and that gives the modifier (so a size 0 creature fighting a size 1 creature gets +10 on some stuff, and -10 on other stuff). As a consequence of this adjustment, my rat now has ability scores that are back within a playable range, and rats interacting with other rats do so the same way as people interacting with other people.

The moral of this story is that edge cases like these need to be tested, games need to be pushed to their extremes to find out where they might break. In this case generating new content based on my rules helped to point out the problem, but rigorous play-testing has caught many other issues as well. Never assume that a rule works just because it makes sense as written.