To open, an amusing comic.
As a philosopher, this question is extremely difficult. In order to make a philosophically sound system that actually has any kind of utility (D&D 3.5 makes me sad or giant frog.) In order to roll your own, I'll suggest a few simple toggles that you can configure for your game.
First toggle: Relativism?
Are these alignments proximate to the characters, or is there a giant beard (or equivalent) in the sky with a bag of presents and a rulebook?
A relativistic, socially constructed, universe (basically all of them without Omniscient beings (omnipotent is fine, so long as they can't catch you...)) has any given morality exposed from the character's point of view, with the external implications of that being an iterated prisoners dilemma. The point of this toggle is the question of punishment: who or what will punish the character if they deviate from their norms?
Example: In the Polythestic monotheism of 3.5 (it makes my head hurt) every character tends to go to the plane of their alignment after death. As K probably says (I don't have a reference) "You only go to the wrong place if you fail at your alignment." Thus, an evil person who performs good acts is sent to, for example, Celestia, and suffers there. A good character who murders is sent to one of the lower planes and suffers there. But an evil person who murders is sent to a plane matching his/her alignment and fits right in.
::sigh::
Anyways, in 3.5 there is a divine source above the gods that notes whether one is good or evil, because of the presence of "Detect Good" and "Detect Evil." If the gods actually powered the spells it would be "Detect Believer" and "Detect opposing philosophies" as most competent old-time religions define evil as "Not us."
Still, the standards of behaviour in a world with a provable beard-in-the-sky will be quite different than one where beard-in-the-sky is a matter of faith.
The toggle resolves to one thing: if the character commits the perfect crime in a forest where no-one knows, could know, or ever find out (see Plato, Ring of Gyges) is the character punished?
If the character is punished only by evidence of her acts, in a relativistic system, then use a reputation system as the basis. If the character is punished by beard-in-sky, then keep a rap-sheet of her actions, with beard-in-sky acting on the rap-sheet in any way the setting deems fit.
Second toggle: Prescriptive and/or Proscriptive?
Does the philosphical system being modeled encourage towards certain acts (prescriptive?), forbid certain acts? (proscriptive?), or judge acts based on their outcomes?
Exempli gratia: in the D&D 3.5 alignment system, if you don't think too hard about it, the Good/Evil axis is prescriptive: I will go [help|murder] that orphan over there. And the Lawful|Chaotic axis is proscriptive: I [will|willnot] break the laws of the society of which I am part. Just don't think too hard about it. It hurts us...
Any given (useful) philosophical system will have both prescriptive and proscriptive elements. However, it's easier to model interactions with toggle one if you restrict most of the interactions to one of the domains.
Third toggle: purpose
There are many possible ethical and moral philosophies. The final question in this simplified system is: is this designed to constrain or guide player action or constrain or guide character action?
Inflicting XP punishments is a way of constraining player action, inflicting in-game consequences is a character constraint. Rewarding bonus XP is a way of guiding player actions, etc...
It is important to be clear about the purpose of your alignment system. From your comment:
"For systems like Hackmaster or D&D where characters have a stated alignment, I'd like a simple system for tracking if the player is actually playing that alignment. In more generic systems which doesn't have a concept of alignment, if could still in some settings be useful to have some metric to evaluate how they measure up in the grand scheme of things. For example, if a daemon can't possess a truly good soul." You want something to track and constrain Player and non-player actions.
I would recommend starting with something taken from the out-of-print RPG Pax Draconis:
Step 1: Have the entity label its philosophy. "Social Darwinism" is one I always cringed at my players taking.
Step 2: have the entity come up with 5-6 "tenets" (prescriptive or proscriptive statements which followers of that philosophy try to obey.)
Step 3: Have the entity choose 3 or more tenets to follow.
Therefore, when distributing rewards, it is possible for you, the DM, to examine if the entity's actions correspond with its stated philosophy.
With this and a simple reputation tracking system with different factions, it should be possible to model any given philosophical framework. This also has the advantage of being an explicit social contract of permissible and impermissible actions on the part of the player.