Rob:
He's had a playtest of it. It's called "T4."
It was, and is, "badly broken." The draft (I've got it on the other 'puters' hard drive) was nothing more (nor less) than a relabeling of the T4.1 task system of 1997-8.
The cure for it is simple, but unwieldy: make attributes cost more to raise than skills, say about 3:1...
CT used a +1, +0, -1 scale for att mods on "tasks" (combat rolls), effectively a 3 point scale covering the 15 point attribute range. The value was about 2-5 reciepts to make the difference.
MT used instead +0-+3 for the same range, but more consistently. The consistency put a 5:1 Att:Skill correspondence. (I prefer 3:1...)
TNE used a 2:1 cost for atts. (It took both hobby reciepts to earn a +1 att.) This made up for the 1:1 ratio used in play. Reduced skills received in later terms of CG.
2300 used un-raisable atts. Problem solved. Range was +0 to +4, in a 5:1 ratio in use. Progressive skill costs, too.
T4 used 1:1 in CG AND Tasks. Even with TIHTIT (This is Harder than I thought), skill reciepts outside of concept are far better spent att-seeking.
GT (& GURPS as a whole) makes attributes about 6:1 for comparable raises, and uses a 1:1 in play. (At least, once you abstract out the various difficulty mods by skill difficulty...) It also uses progressive costs, and doubled attribute costs once play started.
Just for reference: Some other systems correspondences for systems which have been used for Traveller games.
Vampire/WOD: 1:1 in play; 7:1 in CG/XP, progressive costs
Hero: 5 levels of att (5-15 points) grant +1 on skills; +1 on a skill is 2 or 3 points.
CORPS: No difference in costs, but different uses. 4:1 for baseline costs.
EABA: different cost scales. (1d of att ranges from 3 to 18 points in the range of extreme PC's; 1d of att ranges from 5 to 20 points)
BRP - based upon RQ3, range from 2:1 to 1:2, by specific skilll group, on a percentile system. Raising atts not handled in same manner as skills.
Space Opera: in-play varies widely; Atts usually add their level, skills add 5-15% per level.