Well, if we make a couple of assumptions, it's easy to swap proficiences into something more useful:
Assumption 1: The attack/defense math is relatively flat.
Assumption 2: Whatever weapons a class can use, it can use equally well for basic attack/defense (barring some very modest specialization and natural adjustments for talent--i.e. ability scores). Classes that aren't fully martial might have to pay (some resource) to add more weapons to the list, but once added, those weapons are up to this baseline.
So as far as all that goes, there would be relatively little difference with a 16 Str Wizard who paid to use a longsword and a 16 Str Fighter who happened to use a longsword. The Fighter might have some class ability or option to get that modest specialization. The Wizard had to pay a bit more to get to this point, and is poorly positioned to expand upon it, but he is certainly competent with the sword now.
Then proficiences are tied not to attack/defense, but those maneuvers they keep talking about. And you have to learn these proficiences by weapon group--either paying, or because your class is martially-oriented and gets a certain amount for free.
Instead of being proficient with longsword (+3 or no penalty or whatever), you might be proficient with disarm (heavy blades, axes), cleave (heavy blades, axes, polearms), parry (heavy blades), fast charge (polearms), and so on. Perhaps some of the proficiency options are limited by weapon groups--you can't get "fast charge" with light blades.
So an ultra-specialized martial character would only buy proficiences in their preferred weapon group. But even so, if he finds himself without his longsword and picking up an axe, he can still hit about as accurately and hard with it (maybe even better if a magical axe)--but doesn't have access to any maneuvers.
For a more complicated version, you might make buying access to a maneuver cheaper for any weapon group after the first. To keep this relatively simple, do limit each maneuver to 3 or 4 weapon groups. Then it takes one pick to buy a maneuver for one group, and another pick to buy it for the rest that apply. For your favorite maneuvers, it's thus fairly cheap to ensure those are always available, versus spending your remaining picks on maneuvers that weren't your first choice anyway.