I've been working on a system to use for some online settings to put on our website. The idea is that they're quick 'one shots', but each setting will use the same rules.

This uses d6's.
There are two attributes (Body and Mind) rated 1-6 (Dice Pool).
Each Attribute has two talents (Power, Quickness and Control, Reaction), rated 6-2 (Target Number).

There are also Abilities (skills) and Academics (think Shadowrun's Knowledges), rated 1-6 (Dice Pool).

Basically, you roll a dice pool and try to hit your Talent (Ex Swords is a 4, Power is a 5: I roll 4d6 trying to hit a 5+).

You only need one success to succeed, so you purchase qualities with your extra successes. Each Ability has a quality listing, rated 5-2. After totalling successes, you spend a die showing a number, or lower, to purchase a quality. 5 is something good, but 2 is something better.

My problem is that this makes Talent more of a factor than Ability, which isn't true (Otherwise, I wouldn't have quit playing the French Horn), or can you argue that although Ability determines the amount of succeeding, talent determines how well you've secceeded?

Fixes:
Talents can to from 6-1 and Abilities also go from 6-1. Roll Body/Mind vs a TN of Talent+Ability. I don't know if I like this fix.

Talents go from 1-6 and are a dice Pool. Abilities go from 6-2 and are a TN. I no longer need Body or Mind and have to rework Initiative, and Weapon damage (Knife used to be 2/power, now it would have to be Power/x TN).

Rather than have Qualties based on TN, base them off successes. Can't do that; Paragon Games' "Finess System" already does that (Which prompted me to get off my ass and make this system a reality b/c it was on the back of my mind, but I was busy making other things).

Should I keep the system the way it is, or fix it?