First observation: transparency. A casual glance at the game or the stats pages will not yield any explanation on how advancement is calculated. I found it in the PDF manual and the patch notes.
Secondly, rank is simply determined by player_hours*effectiveness.
As an incentive, that is insulting. As an aid, it is almost useless, but then, nothing in the game (e.g. matchmaking, team balance) seems to be using the rating/ranking anyway?
Somebody coined the phrase "benchmark for skill" to describe what attracted him to SH. It's a good selling point, and a worthy objective.
But that means you need a good skill estimate. Such estimates exist:
The basic idea: your existing rating (skill estimate) will predict the likely outcome of a match against any opponent. The more certain the estimates, and the larger the difference in skill estimate, the less surprising the outcome should be. Surprising outcomes make large corrections to skill estimate, expected outcomes make for small or no corrections. The more often an estimate has been corrected, against well-estimated players, the less it is seen as uncertain.
Now obviously, crowded MP matches do not make for clean pairwise comparisons. SH does not have a tourney mode (why not? - it certainly has servers to spare). So estimates are going to be messy.
That's OK - if you really need a number to tell you how good you are, that does not invalidate the basic idea. Usually, there is a lot of debate about the details, but the math is straightforward, and you can do skill estimates without throwing away straightforward stats and counting.
Now let's say you have a small user base that plays frequently, and you can accumulate a lot of data of who scores and caps, who damages whom, who kills whom, and how effectively and quickly. If need be, take into account how crowded or empty the server.
I.e., just for discussions sake, let's assume it is possible to estimate skill in some meaningful way under less-than-perfect circumstances, and get somewhat useful results.
What then? Matchmaking, team balancing, handicaps.
Or, most interestingly - move away from counting frags, and show who gained and lost most in the skill estimates. Why? Because now it pays off to go after the stronger players, instead of preying on the weak. If the skill estimates are essentially worthless - or if individual skill varies a lot from map to map and mode to mode - you will see wild swings instead of steady convergence. But the basics - movement, aim etc. - should show.
On the other hand, you definitely want to learn from Halo2/3/Reach and make sure that matchmaking is guided by skill estimate last, and desired map, mode, group, friend etc. first.
If you want to push this all the way, make the HUD markers indicate risk/payoff: the more of a proven threat a player, the more he stands out.
Maybe the prospect of collecting match points is what keeps you playing, but as a benchmark for skill this is completely useless. We can already count hours played, and points/match. I'd be much more interested in something that tells me about raw ability, no matter how few hours were spent on the game.
Right now, we seem to have only the "Experience" part, not the "Skill" part, of the Halo3. Granted, even Bungie is still fiddling with their system
But however you use it, having a skill estimate as part of your stats is crucial if you want to be seen as a "benchmark of skill". What we have is epxerience-only:
The aim of the new ranking system is to provide better rewards for all players from people new to the game through to our seasoned veterans. The new system uses the same 10 rank titles from Recruit through to Commander but each title now has sub-ranks named Iron, Bronze, Silver, Gold and Platinum. This system creates 50 ranks in total allowing lower ranks to be grouped closer together and experienced ranks to be spaced further apart.
I for one would prefer a system that uses the colors for experience, and the ranks for skill, instead of introducing sub-ranks for experience and no score at all for raw skill independent of hours played.