It has been discussed to death, but ultimately it's just not a good idea. Many builds have virtually no weaknesses aside from balagarns for being dex based(OH DEAR I DO 10 LESS DAMAGE NOES!!!!!!).
Among them are: Bards, DWD's, and Harper Scouts. In the end ALL that really happens here is other dexers become good, but still nowhere near the power level of the uberness of the dexer equivalents who's only real weakness WAS balg and in addition because these dexer uber equivalents now lack any weaknesses and now most dexers are fairly effective... It's a hard pulled reason to build str based at all aside from RDD. Also Balagarn's Iron Horn is hardly a hard counter, because both the caster and defender roll a d20 against it, any point of str score above 20 reduces the chances of balg by -more- then 5%(approximately 7.5% each one).
A good example is 30 str, a dexer with 30 str means the caster needs to roll a roll of AT LEAST 11 and the defender has to roll a 1
The way odds of multiple probabilities works in statistics, is the probabilities are combined into one singular asymmetrical probability, this means by definition if both the rolls have the same base, the probability is simply 20 of 40(40 is the max from both of them being 20 and 20 is the highest any one of them can be), which means by definition if a build has 20 str it -automatically- gets a 50% chance to resist balg. Any number(recall it's str score, not mod so odd numbers count too) above/below 20 changes the random by approximately 2.5%. A str of 30 means it's 30/40 = 75% chance of resisting, now I bet you're wondering why I don't make the maximum possible 50, well that's because the 10 being added isn't a probability, it's a direct number addition on top(The 30 str score isn't random). By the same token the chance of balg failing becomes 10/40 = 25%. Now what happens if str is below 20? Simple!!!! The same occurs except now the maximum for the resisting side drops to 18/40 = 45% and the casting side increases to 22/40 = 55%
You'll notice NONE of my dexers are easily balagarned, well this is BECAUSE I take bonus strength on ALL of them, I sacrifice about 1 ab/AC for 2 disc, 2 damage, more carrying capacity and a HUGE decrease to balg(generally either 13 or 14 str, though if I can I take 15 because odd numbers still help).
Well EVEN 13 STR helps immensely, it's 25, which is 25/40 = 37.5% chance of failure.
NOW What I could possibly do is reduce the roll of 20 for the caster to a roll of 12 and in exchange double the duration of the actual kd. As it is now, if you can't -chain- the balg kd's endlessly it tends to very rarely result in death unless the guy was all alone and gets ganked to hell or he was already about dead, how's that sound? A roll of 12 instead of 20 would make a build with 26 str only have a 26/32 = 18.75% chance of failure, which is still possible to balg, but much, much lower.