Bounce this dice mechanic off your heads

Hello all,

As I am currently stalled out with my other game (waiting to find playtesters) I am working on a new game system with a fairly common design: I want to be able to plug and play different conflict resolution rules and reward systems to create three (or more) different systems within the system in order to be able to encourage all of the creative agendas (on their own, not at the same time). At some point I may experiment with combining techniques and rules to try to encourage play that focuses on "hybrids" but for now it is not the main concern.

To this end I have cooked up a fairly simple dice mechanic to use within conflict resolution and want you guys to weigh in and bounce it off your heads so I can develop it further and make it interesting.

Basically, characters can have (depends on how the system is constructed, sort of like Fudge) Attributes, Skills, Traits, and all kinds of other things. No matter what, each one will have a value, and that value will correspond to how many dice it is worth.

So far, when a conflict is entered, you roll the appropriate attribute dice only (and maybe skill dice, if applicable) As all dice are d6's (something I am toying with changing, for reasons that should soon become clear), the results of each die roll falls between 1 and 6. My dilemma is what to do with these results from here.

One thing I am thinking of is that each result is "spent" in a different way, as such:
Spend a 1: Something happens
Spend a 2: something different and better happens
3: something still different and better happens
4: something still different and better happens
5: still different and better
6: roll one of your traits and add its dice to your available pool of results

This system would not allow you to add results together, so you can't take a four and a two and spend them to roll a trait. It also assumes that when you make an "attack" on your turn, yu can spend as many or as few of the dice as you like and total their result. So, a six becomes valuable because you can either potentially get more dice or have a very nice attack value for a minimum of dice invested. The problem is that I see spending a six to roll a trait to be a better buy all of the time, since as long as you have a 2 or higher in the trait, you will likely beat the six total anyway. One fix I have thought of for this is to say that when defending against an attack, you cannot use more dice than the attacking player used, which makes a six significantly stronger.

The other idea I am toying around with is having a list of options, maybe 4-10 in which you can (and have to for the higher costing options) combine die values and spend them. This way I could make rolling a trait cost 10, also fixing the above problem of sixes being too good for attacking with.

Alternatively, I could operate as I did in the first mechanic, but have values greater than six, so that the only way to access them was to use bigger dice, say a d8 or a d10. I could assign d8's to skills, and d10s to traits and things, leaving statistics at d6. Since the only way to roll a trait might be to sacrifice an 8, which is rare since skills will have lower stat numbers than attributes, it would make it difficult to escalate up. But it might cause the same problem, where sacrificing an 8 doesn't mean much if it lets you roll 2-4 10 sided dice.

So, what do yo guys think? Which dice mechanic seems the most intuitive? Which encourages tactical thinking and which doesn't? What changes would you make to ensure the choice between spending for more rolls and attack values is hard? What other thoughts does it provoke?

Cheers!
--Norm