Sounds cynical, I know. But here, walk with me:

1) Backintheday, music was distributed on vinyl. Vinyl has some distinct advantages: a good disc on a good deck through a good system sounds really good, whereas on a CD, pretty much everything in the digital stage is identically mediocre. Vinyl also has some distinct disadvantages: pan the bass too far to one side of the stereo field and your record will skip. Use a dull lathe and it sounds like ass. Exceed the RIAA profile for your equalization and introduce artifacts. So "mastering" was the art of getting your music from "studio format" to "consumer format" where consumer format was transistor radios, 8-tracks and Montgomery Ward console entertainment systems. It was basically a way to get your music into "the real world" and reflected some "optimization" of the full spectrum of sound (multiband compression, soft-knee limiting, RIAA EQ, etc) in order to use every part of the dynamic range available to you.

2) When Sony introduced the Compact Disc, every playback system was as good as every other, and that playback system's baseline was right about at "reasonably good vinyl setup". There were a few hiccups while engineers around the world learned the wonders of dither, but by and large, 16-bit, 44.1kHz digital is a big step up from K-Tel. So now everything sounds pretty good and everything has pretty good dynamic range, but we still want it to "pop" on the radio and MTV, because how else are we going to sell the disc? So "RIAA curve" fell by the wayside and "maximize loudness" became the main principle.

3) Every single one of us now listens to MP3s. Our standard playback is tiny, shitty earbuds and a 320kHz lossy bitstream. The DACs on iPods and iPhones have sucked pretty hard since '08 (whenever they rev'd the last hard disk-based player) but let's be honest - they're taking the place of cassette tapes. And those shitty little headphones are orders of magnitude better than non-shitty headphones from even 10 years ago, thanks to the development of really good, really cheap neodymium magnets. Either way, the art of taking a bitstream lossy is its own "mastering process" and we have much less control over it than we really need (thank you Fraunhaufer) and it has not a lot to do with making the music fit into a given dynamic range and quite a lot to do with making it fit into a given bandwidth.

In other words, the reasons we came up with "mastering" are long gone. Even the last holdout - "mastering makes all the tracks on an album sound similar to each other to improve the album-listening experience" - is completely nonsensical in the modern world. Most people no longer listen to albums, and most bands that care about albums record them themselves. The era of an album being cut in six different studios by nine different engineers is over and it is never, ever coming back.

So, that's why "mastering" exists. Now, three stories:

1) A friend of mine wanted to be an engineer - we'll call him Bob. He was a MicroSerf; he could just go out and buy $40k worth of gear without thinking about it. I took Bob to an AES meeting about "mastering" in which we listened to three qualified mastering engineers butcher the shit out of someone else's material and he said "shit, I can do that." Found out later Bob had actually signed one of my other friends (not a mutual friend - we'll call him "Bill") to "master" Bill's album (http://www.myspace.com/smphq/music/albums/hacked-10390009) without really knowing what to do. Bob had me over and I explained the "principles" of mastering and let him be. Bill later told me that Bob had done the best mastering job he'd ever heard. When I told Bob this, Bob laughed - he'd tried a few things, he'd bought $2k worth of plugins, and in the end, he just spent nine hours drawing fader rides in Logic.

2) I went to an AMPAS event once where Bernie Grundman was speaking. Bernie has been one of the go-to guys for mastering for about 40 years now. And he's talking about theory, and he's talking about practice, and he's talking about all the subtle, beautiful things mastering can do, and then he says "and here's a great example" and proceeds to play Michael Jackson's "Rock with You" off of "Off the Wall." And it was beautiful - light, airy, almost jazzy. I was ready to eat my words and think "okay, when people really know what their doing - " when Bernie interrupted my thoughts like the record scratch in Allie McBeal with "and here's what it sounds like when it's been mastered: http://www.youtube.com/watch?v=z040cAt98zE And I thought "yeah. There it is. Smashed to within an inch of its life, and nobody even noticed."

3) I'm "mastering" a movie soundtrack right now. I mix it, get it within the limits, and then pass it through some heinous, monstrous preset built into Logic. That preset has linear phase EQ, multiband compression, multiband limiting and a brick wall limiter in it and their "final movie mix" is fuckin' aggressive. I took out the EQ, reset the multiband limiting and use the brick wall limiter as a clip gauge - if I hit it, I change the material. See, I have to hit DIALNORM and I can't pass +9 (ref -20) or the mix gets rejected by distributors. That's a broadcast thing. But more importantly, it's a supacheap zombie film and the guns need to sound like guns while allowing the movie to sound like a movie. Even with that, though, the best way to "master" it is to take out half the bullshit that Dolby thought it needed by default.

"Mastering" is largely bullshit. You don't need it. Don't try.