FPS game developers are going in the wrong direction when it comes to multiplayer. I want another awesome 1v1 game!

Adding classes, achievements, unlocks, or whatevers is not the solution. The real key to depth is having systems that people are able to master within the scope of the fps skillset.

Add movement systems similar to QW where your skill at bunny hopping has a pretty direct correlation to how fast you move. This is an example of something that is skill based and has infinite degrees of mastery because the closer you get to the sweet spot of accelerating with a dynamic movement of the mouse, and the more consistent you are, the faster you can get going. The most ingenious thing about bunny hopping is that it is like a "powerup" but exists completely in the FPS skillspace. If you have the skill, you get the movement "powerup." And if you have more skill, your "powerup" is stronger. It is pure genius (too bad it was an accident).

Think of the things that amaze you when you watch a frag video. Crazy jumps that you never thought possible, mid air shots, twitch rails. These are the concepts that need to be expanded upon and perhaps somehow reflected in the scoring system (like 2 points for a midair, etc.). I'm sure others can think of more examples of these. The development that needs to be made is fully embracing that these are the elements of the fps skillset and have them be completely apparent as part of the gameplay system (with documentation and tutorials). In the RTS genre, this has already been done with full documentation of skill trees and damage types, so that everyone knows the rules. The skill comes in understanding when to build what and how to use it to counter your opponent.

If map, physics, and weapon design account for these gameplay mechanics to be taken to the extreme, the depth of the game will be huge. I think this has been done fairly well in certain fps games, but the problem has been that these gameplay elements are not made apparent enough via visual cues, scoring, sounds, hud elements, or whatever else. Developers have the right idea, with the badges in Q3 (or whatever they are called), for pummeling, getting rails, etc. Achievements in TF2 attempt the same thing. The problem is that in-game awards such as these, up to this point, have not been sufficient for new players to understand why they are being owned.

If Quake3 said "ARMOR RUN" (but some sort fancied up wording) and displayed a badge on the screen every time someone strafe jumped flawlessly from ra to ya on dm6 and got both armors, or it calculated how fast you were moving your mouse, or how fast you were moving when you hit a rail shot. Or if you dodged some rockets successfully. There are many examples. If these could somehow be incorporated into the scoring mechanic (not necessarily an annoying voice saying "ARMOR RUN"), the "depth" of the game that is apparent to the average person will be greatly increased. It just comes down to quantifying all these elements, and creating a system to make it apparent that a players' use of these elements is why they are better than you.

The other HUGE thing is lack of matchmaking. I think if they add this in Quake 5 it will be a giant step forward. The reason StarCraft has become so successful is that it has a built in system to make sure that new players have a positive experience. You will get matched up with someone equally bad, you will lose a match, but it will be very exciting and close, and there will be some small quantifiable reason why you lost. You will say, "oh when he did this I should have done that, next time I will do that." Then after a few matches you start playing better players and the process of refining your skills and gaining experience continues. In fps games, you come in and lose 60 to -10 five times in a row and have no idea what to do next. Then you stop playing online, and go back to playing bots. Then you beat the nightmare bot and you have "beat the game" and you go buy something else.

If you always had the chance to play people at your skill level, your matches would be close enough that you would usually be able to pick up on something that went wrong, and you could make incremental improvements. People need to understand what it takes to play the game well, and the learning process needs to be fun. This may also solve the problem of everyone being so stubborn about staying with the game they originally loved (qw, vq3, cpma, etc.), and being able to master something new, since they will have a framework to build their skills without instantly being owned and then longing for the good ol' days of [insert name of game that less than 50 people play].

Recently the industry has been going in the wrong direction to make FPS games "fun." An example would be TF2, which is so shallow and leaves so little room for skill, that any new player can hop in a server and not get completely owned from the start. It seems like this solution really encounters the same problem. Instead of losing newer players when they get discouraged, it loses experienced players because they get bored. Then they put another bandaid on the problem - unlockables - but they fail to realize that grinding out achievements isn't really a form of "being good" in the fps playfield, and so does not remedy the depth problem. If they had designed TF2 to have a deeper gameplay experience, and added matchmaking, it would have solved both problems.

So just follow my advice:
a) include gameplay elements that account for and allow infinite skill, and scale in such a way that the rewards of mastering these elements are directly correlated with the players' actual skill.
b) quantify those gameplay elements in a way that is obvious to users
c) include matchmaking

...then incorporate all the advanced competition features of cpm or kteams at launch, and you will have the most successful fps game of all time!