I talked a bit yesterday about how pleased I was with the Dark Sun monster book because it had the tools of play baked right in. That got me thinking a bit about monster books in general and how well (or poorly) they line up with how they actually get used. (This is all pretty 4e centric, but theoretically applies to any game with a bestiary.)
For me, monsters fall into two categories - reliable and stunt. Reliable monsters are ones that I'll use a lot in any given game. Kobolds. Skeletons. Orcs. Creatures which I imagine existing in the setting as "monster races" or otherwise having a reason to exist in large, interchangeable numbers. 4e proved an absolute delight for these guys since it embraced the idea that there might be lots of creatures of that type representing a range of threats. That made them a lot more useful to me.
Reliable monsters also have a substantial impact on the setting of the game because they don't come from nowhere. The goblins live somewhere, and the undead were once not quite so dead. They become part of the fingerprint of the setting.
Stunt monsters are pretty much everything else. I pick them up to jazz up fights or to build fights around. They're one-offs, and while some of them fit into the setting in a high level sense, there's not necessarily a huge need to put a lot of effort into it. Ultimately, nobody really cares where that gibbering mouther comes from once it's been looted and the party has moved onto the next room.
Now, it's possible this is entirely idiosyncratic on my part, but I think not. Consider the way adventures make use of new monsters - it is a rare adventure which does not have at least a few unique monsters which are unlikely to ever bee seen outside of the bounds of that adventure (and maybe DDI). These monsters serve some particular purpose in the adventure, and that's much of the fun of them.
All of which is to say that this speaks to the importance of putting plot hooks right into the monsters. With reliable monsters it might be important to give some amount of background and social context, but for the bulk of monsters, it seems the thing to provide is guidance for how to _make_ it that interesting one-off.
Curiously, one of the most interesting examples of how to do this can probably be found in people's write ups for their own version of New Crobuzon. Since the basis of the writeup was (effectively) pick three reliable and three stunt monsters for your city, you get to see some really fascinating ways to handle both types. Because the stunt monsters are explicitly designed as one-offs, they get writeups that are all about play, not about bad biology essays.
1 - Birthright did this quite cleverly with their "Orogs" which basically were a single bucket into which a number of evil humanoid races were tossed, with the idea that you could run into small runty ones or huge, ogre sized one, but they were still Orogs. Given that they had a national presence in the game, this worked surprisingly well, though it only came up every so often - Birthright was usually about people and Big M Monsters. Handling Orogs this way felt suitably Tolkeinesque.
2 - There's an exception to this in the form of Big-M-Monsters, monsters that are big enough that as individuals that they are elements of the setting. Named dragons are probably the best example of this, but things like the Beholder crime boss of Waterdeep count too.
3 - If you haven't read these you absolutely should. They're some of the best examples of how to take a small set of 4e elements and make something fantastic. I admit, I'm totally proud of mine, Vicidia.