What will probably not be a well received idea...

[RobbyPants]

This type of system tend to favor the players, as they're expected to get into fights against weaker opponents most of the time. This isn't bad, but something to keep in mind.

[nijineko]

how will you avoid the trap of replicating extreme specialization without sacrificing points needed for normal / effective defenses?

i have observed that in essence, class levels (with associated abilities), feats, powers, and spells are in effect predetermined bundles of points used to build and accomplish some goal, or bring some ability to the table.

how will you decompile the various d20 abilities into your point buy system?

[bhu]

Isn't really point buy so much.  There will be a set group of abilities you can choose at 1st Level, and occasionally at higher levels.  You can choose 2-4 of those abilities at 1st Level, and with each level you can upgrade one.  For example a typical 1st level spell

Burning Touch
Evocation [Fire]
Level:    Fire 1, Sor/Wiz 1
Components:    V, S, M
Casting Time:    1 standard action
Range:    Melee
Duration:    Instantaneous
Saving Throw:    Reflex for half
Spell Resistance:    Yes

You do x additional fire damage with a successful melee attack.

For upgrades you could shorten the casting time, get rid of components, extend the range, increase the duration or damage, get rid of saving throw/spell resistance, add additional effects, make it a touch attack, etc.

Fighters will be basic ToB style maneuvers that are modified the same way.

[nijineko]

that sorta sounds like what they were trying to implement with 4th. the concept anyway. i see some differences.

[RobbyPants]

For upgrades you could shorten the casting time, get rid of components, extend the range, increase the duration or damage, get rid of saving throw/spell resistance, add additional effects, make it a touch attack, etc.

I had a similar concept a while back, but I never did anything more than think about it.

One pitfall I can warn you about is that this will play out similar to metamagic feats. Any one given MM feat is usually okay or crappy for most spells, but will interact differently on a spell by spell basis. Certain combinations simply work much better than others. That being said, you will have to take care to figure out on any given spell/maneuver/power how it interacts with getting a shorter activation time vs getting a bonus to the effect. Often times, one will be much better mathematically.

Also, you'll likely want to adhere to some sort of double-a-double-is-a-triple rule to keep things from getting out of hand when you apply multiple enhancements. Or you can go all Diablo 3 and only allow one enhancement to an ability at a time (like their rune system). It's crude, but it does cut down on unexpected super-combos.

[nijineko]

i was kicking around an idea for a magic system where each characteristic of a spell has a unit cost, with more units having greater effect within that characteristic.

the end "level" of the spell is determined by the total cost of the spell, which level in turn caps the maximum limit of each of the characteristics. casting time would also be determined by level.

metamagic would simply become the ability to tweak the unit distribution of a spell.

wizards would use "pre-calculated" spells and thus cast faster, while sorcerers would have to cobble together the spell effect each time, but would have innate and ultimate flexibility.

i also thought of splitting up types of magic based on the epic seeds / magical concepts, with access to various kinds and types of magic based on class and level.