What I would love to see is something like the 'Master of Arms' supplement from Second World Simulations, which came out early in 3e.
It essentially used combinations of base and secondary (or even tertiary) attacks to pull off some excellent manoeuvres - and it provided the maths to back them up and help you develop your own if you wanted to.
e.g. Axe Beheading (battleax, greataxe, waraxe). Sacrifice the first attack to increase your base critical rate to 18-20 with the axe (i.e. you take only your secondary attack).
e.g. Blinding Slash (slashing weapon). Two step combination, you first make a targeting attack against the victims normal AC which doesn't do any damage. if the targeting succeeds you then follow with another normal attack which blinds the target for 1d6 rounds if Fort save failed.
e.g. Double Spin (bladed polearm: glaive, guisarme, halberd). You spin the polearm high on the first blow and follow up low at the legs. You take a -2 to your first attack in the sequence but +3 to hit on the second attack in the sequence.
The whole book has got lots more examples - I think it would be a great way of providing a huge, flexible range of martial manoeuvres (it also includes cloak fighting, three section staff fighting, immovable rod fighting and others!)
As you can tell... I'm a fan.
Cheers
The trouble with this and non-damaging attacks is that a person has to see more value in spending an attack that does no or less damage compared to simply smacking the creature.
Often the alternative that is more complicated needs to be a recognizably larger advantage or the person/player will select the easier option which is simply to hit the creature for strait damage. (this larger advantage when spotted then becomes often the focus of exploits though it is a 'reward' for the person that did not choose the simpler design mold).
This is a design problem where you can make adoption of the idea easier if you get damage and the extra rider.
The trouble of moving into a rider/proc system is that they add more complexity to each turn's resolution and can slow down play. This is again a trade off on value of quick resolution (hit and hit it more) verses keeping track of the effects of an attack that slightly stunned an opponent so their attacks are weakened (doing less damage per hit as a result with the risk of forgetting the effect on the target).
I like the concept but it is a tricky design goal to achieve with many associated problems.