So you admit that your pool too favors faster miners when you specifically (incorrectly in every way except your one shortsighted emotions over statistics way) chided Slush's pool for doing so in your OP.
"Testing with" and "favoring" are two different things. We pay CPUs and GPUs the same amount for each share. We don't "favor" either one over the other.
Your pool asks the miners to keep a high Share/GetWork ratio. In order to accomplish this, your pool requests that miners process the entire solution space of every single GetWork before requesting another GetWork. This means that the slower a miner, the more likely that that miner's solutions will be Stale.
True. So if the CPU user's can't handle that, then it's time for them to upgrade to a $89 video card and get with the times.
Simple as that.