In the earlier days of the hobby, most of the games that people played the most came from TSR--the producer of Dungeons & Dragons. This isn't surprising. What is perhaps more surprising is that none of these games really made an attempt to capitalize on the familiarity of using a similar system to D&D--games like Metamorphosis Alpha, Gamma World, Star Frontiers, Gangbusters, Top Secret or Boot Hill tapped all kinds of genres--most of them pulpish action genres that had at least enough similarity in tone and feel that they could have benefitted from a similar approach. But they didn't really; they were all completely stand-alone games.
After D&D started permeating the market a bit and competitors started cropping up, some other publishers used "house systems" for their games. These weren't exactly meant to be generic, but they were meant to utilize the same system, tweaked and slightly modified for the specific game that they were being used in, but otherwise operate very nearly the same. The Hero System, first published with the superhero game Champsions was one early one that later spawned Star Hero, Pulp Hero, Fantasy Hero and more; gradually evolving into a generic by design system. The Basic Roleplaying System had a similar provenance; it was the system of Call of Cthulhu, RuneQuest, Stormbringer, and many others... it later evolved (in fact, it was probably among the first to do so) into a generic engine that is expected to apply broadly to a wide variety of settings and genres with minimal tweaking. GURPS was probably the first to actively pursue such a goal; published as a generic system (the G in GURPS is for generic) with setting-specific add-on modules. More recent such engines, which are specifically designed to be generic, include Savage Worlds, FUDGE, FATE, and d20 Modern--a derivative of the 3.5 rules of D&D. Others, such as Storyteller, CODA, the d6 System, or Unisystem, have never tried overtly to be generic, but are included as the "house system" of a publishing house, and as such they find their way into a variety of genres and can readily be adopted as nearly generic systems as well.
Why do I like generic systems? Why not one that's designed specifically for the needs of a given game? Mostly because I don't really like system for its own sake anymore. These days, I prefer the ease of slipping into a system that's well-known and which I can play without having to worry about, "do I and my players actually know the rules well enough to do this without it being a hassle?" One way to do that is to focus on well-known systems that can be easily adapted into a variety of genres. This is exactly why--although it's not really my Holy Grail of system designs by a long-shot--I've decided that d20 works for me. I don't have to worry about knowing the system. I usually don't have to worry about my players understanding the system, since almost everybody in the hobby has played a d20 game at some point or 'nother. The trick, then, is finding the system to settle on, and for that, you want to make sure that the tone that the system fosters and the one you want at the table are in synch. That's exactly why I'd never really entertain adopting GURPS, or Hero or Rolemaster as my basic system--all focus on minutia which doesn't cater to the kind of tone and feel that I want in the game. I prefer a much faster and looser approach.
I also like relatively durable characters that I can throw stuff at and not worry too much that they're going to die willy-nillly. And finally, I like to have a lot of material available. I've said it before and I'll say it again: it's better to have a tool you don't need than to need a tool you don't have. The biggest reason I prefer d20 to any other generic system--aside from the ease of finding players who are familiar with it already--is that I have so much d20 material that I can use. I have so many monsters. I have so many modular rules add-ons. I can tweak and mangle the rules easily--usually without even having to do any amateur development of my own, because I can easily find somewhere in print where someone's already done it.
It's always surprising to me that d20 Modern was specifically billed and presented as a modular, generic system, whereas D&D 3.5 is believed by a sizeable chunk of the market to be a carefully balanced, tightly integrated system where nothing can be changed without risking the entire thing toppling over in disaster--considering that the two games are virtually identical system-wise. I've always found that d20--of any variety--is almost painfully easy to houserule and modify, to the point where I can't really even remember playing it completely as written pretty much ever. And the fact that there are so many options and variants in print--to say nothing of the house-rules of gamers all across the world--should belie the notion that it can't be done easily enough. But it doesn't seem to. I really don't get it.
To me, a good generic system should probably leave F/X--magic and whatnot--to modular add-ons. The magic system for D&D is one of the main culprits in terms of making D&D non-generic. The non-magical aspects of the system, however, can and should be generic enough that they'll work for a variety of settings without change. In this regard, again, D&D fails because so few of the core classes (archetypes) are non-magical that you have very little you can do with character generation that doesn't involve the very specific D&D magic system. d20 Modern on the other hand, is an action-hero system that can produce all kinds of non-magical characters with a lot of variety--and then F/X is given to us as three sample modular add-ons that work differently to encourage slightly different expressions of tone and genre.
I could have also used d20 Past; in fact, for a long time I did overtly do so. However, since that was never part of the MSRD and its now out of print, I decided to go for a simpler model that just ignored it. There are a few Occupations and feats that would be great to use in DARK•HERITAGE for those so inclined who have the book; but otherwise, you hardly need it to proceed.