Hm.
I think I've done pretty well, honestly, and I find it a little offensive to be painted as a sort of ignorant fool that doesn't 'really' understand the games I run.
It's really not that hard. Rules comprehension is a skill you can develop. The rules are not a black box, only revealing their effects through careful study during play. They're all laid out in the text in front of you. Some games (again, Exalted comes to mind) have everything so interlinked that it IS really a pain to work out the ramifications of even a minor change, but such is not always the case. Many games are fairly straightforward or modular in design, such that you can modify one part or another without doing any harm. Yes, there's a chance a house rule will be a mistake. But you don't need hours of playtesting and empirical evidence to realize that someone made a math error or didn't consider an important edge case when they wrote a rule. To use an extreme example, I really, really don't need to experience 'bucket healing' in play to decide that it's stupid and declare that drowning does not work that way.
In fact, sometimes empirical experience is actively a bad way to develop a houserule, especially when you're working with probabilities, as you often are in an RPG. Personal experience is notoriously susceptible to bias and superstition when dice are involved, while it's relatively easy to apply statistics and thought experiments to map out the results of a change.
A lot of game systems just have bad math, incomplete-but-interesting subsystems that could use a little more detail, and, often, contradictions that outright need house ruling because the RAW are incomprehensible.
'Bucket healing', by the way, came up earlier in the thread; it's a weird obvious loophole in the d20 drowning rules that lets you heal a dying (between -1 and -10 hp) character up to zero hit points by briefly drowning him, such as by dunking his head in a bucket of water for a short time.
