The original post is a bit misleading.
Now, I don't know how the bash formula exactly works, but from my personal experience, I've had a giant bash me (halfling) and sometimes the bash lasted four rounds and a half (VERY close to five) and sometimes it lasted one round (note that I have no idea whether they ate a venom sack).
Of course the cases where it lasted one round are much less than the cases where it lasted almost 5 rounds.
From the huge lag difference I have experienced, I believe there are more variables that define the lag you get from a successful bash. I actually believe that the bash formula is quite complex and there are at least a couple more variables than just size and AC. Actually, from a recent discussion it seems like being mounted also gives an advantage in bash avoidance.
No matter the variables though, being bashed and getting ~5 rounds lag from it is ridiculous. I am strongly against capping PC sizes up to giant, but I am FOR changing the way bash formula works. One should no be able to be bashed for more than 3 rounds. There should also be diminishing returns from stacking bashes. That does not mean bash should have more chance to miss, but it could mean less lag.
Also, no matter how much people claim get AC and you will be safe, that is not the case really. Getting -400 AC is not easy as a melee/rouge/hybrid, even with -100 AC from buffs alone. Hell even -300 is not easy.
That being said, when my DECKED halfling blademaster died to Grum, he was bashed 4 ticks in a row without him missing a single bash. It is funny, because his DIRT was missing 50% of the time while his bash was landing with surprising efficiency.
Just my 2 cents.