That's still a pretty big error. It only work out correctly if ALL modifiers are either attack or defense mods.
It depends on what you mean by "correctly". This is the way that Civ4 has always done it and it was apparently done on purpose. But BtS does not normally get anywhere near the modifier totals that C2C can, which increases the difference between what it does and what you might expect.
So it is true that is is not giving the same results as applying all the attackers bonuses to the attacker an all the defenders bonuses to the defender, or the equivalent of applying them both to the defender but independently (as Dstr * ( 1 + Dmod) / (1 + Amod)). But this does not make it an error if it was what was intended.
If only one of the two gets a modifier it is clearly the same in either case. It only matters when both get modifiers (and the attacker's modifiers are not the one type that get applied to the attacker). In your example it doesn't matter which way the 100% bonus is applied since 6 vs 1000 and 3 vs 500 are equivalent, they are both 166 2/3 to 1 ratios in favor of the strong one (leaving the weak one with identical practically 0 chances of winning in either case).
You might want to see how a few other cases work out to see why it was done that way. What effect would it have to "correct" the calculations vs. the way it is done now for some city raider macemen attacking city defense archers with a few variations like the macemen having one more promotion than the defenders, or two more, or one less, and the city still having 10% defense or being on a hill (which is a big bonus for the archers) in each case?
Two basic examples:
Assume both attacker and defender are strength 10.
1) Defender has 10% more bonus than attacker, with 30% vs. 20%
10 * 1.3 / 1.2 = 10.83 defender strength ("corrected" way)
vs.
10 * 1.1 = 11 defender strength (existing way)
So in this case the existing way is better for the defender.
2) Attacker has 10% more bonus than defender, also 30% vs. 20%:
10 * 1.2 / 1.3 = 9.23 defender strength ("corrected" way)
vs
10 / 1.1 = 9.09 defender strength (existing way)
So in this case the existing way is better for the attacker.
Based on these two examples (so not exactly a complete examination), it looks like for the existing way actually magnifies the benefits for the one with the higher bonus, at least for these sorts of cases.
Since in BtS the defender often has higher bonuses due to terrain, features, and fortification it looks to me like it is set up so that the defender will normally have a small increase to their advantage from this. But if you can get your attacker promoted enough to get a higher modifier than the defender's then the modifier advantage will flip to being magnified for the attacker.