Honestly, there is only so far that theory-hammer will get you. You know that things like redundancy and mobility are important. You can do some math-hammer in your head. However, you need to play games in order to really see concepts in action, to see them succeed or fail.
Playing the game of course involves an army list. And here lies my point: are you one of the folks that aims to have a single army list, or do you switch among several?
The Single Army List
Technically, I have a couple of 'single' armies right now. My Tau army is a Stelek army. My Black Templar are an adaptation of a Stelek army (Credit where Credit is due). The BT are also build so because I had a bunch of Termies laying around from Deathwing days, but no urge to play with the DA codex.
The goal of the single army list is, well, to get one generalized all-comers-capable list. The Tau are prone to this because the codex doesn't have a lot of options. The Black Templar are this way for me because I have no great urge to re-start a marine army.
The basic pro of the single army list for a codex is that you stop worrying about building the army and can focus solely on using your army. I've played a decent number of games with them, and most of the time after the game I'm asking myself 'How could I have used my units better' instead of 'how could I change my army build?'
In a sense, this is the major benefit of the single army list. Once you have decided on the list, you are free from worrying about changes to the list. It is a way to cut down the brainpower associated with the after-action analysis. It also increases the benefits of practice: you will gain a better idea of what each unit can do on its own and in the context of the army list as a whole.
Economically speaking, the benefit of a single army list is that you have fewer models to buy. You buy everything, paint it, acquire foam trays (or whatever your storage/transportation solution is) and you are done with the economics and logistics consideration.
Multiple Lists or the Evolving List
This approach involves changing up your list. In a sense, everyone will start here at one time or another unless you pull a solid-looking list off the 'net. You'll start with the basics, expand to a full range of models, and experiment with your list. You'll change based on your fancy.
The bright side? You start to get an appreciation for the various units of the codex. You can learn and adapt. You can take a fresh list, and take a break from another list you run. This is what I'd done with my Eldar before I started scrapping builds bit by bit, and settled on a small core (or, more accurately, got fed up with the lack of options). This allows you to own a lot of the cool stuff in the codex, even if you couldn't conceivably use it all the time.
You'll probably have a core you start from; at the minimum it'll be a couple troops units. You might change up the supporting elements, increase the number of troops choices and change, or whatnot. It's kind of an organic expansion; you decide this will be neat so you proxy it, run it, decide it works, and buy it. Once you have enough stuff, you can change your list from week to week.
Economically speaking, you're always going to have some surplus units. Why? Because your unit stock rotates. One game you might use a unit, another game it might sit out. Some stuff might get used all the time; this is usually a couple of troops units and HQs.
Some chassis might be useful all the time, but in different configurations. Marines are the prime example for configurable chassis; the Rhino body covers so many designs. I mean, with magnets you can make a Rhino chassis double as a Predator, or you can simply replace the top to make it pull double-duty as a Razorback. Similarly, a Land Speeder can drastically change roles from, say, a multi-melta/heavy flamer suicide job to a heavy bolter/typhoon fire support model.
Sometimes, it's just a squad with slightly different loadouts. You'll always want some bolter marines, but sometimes you want to switch special and/or heavy weapons, or the sergeant's kit. In this case, you'll have some spare models on a squad level, but you'll generally get a lot of use out of the basic grunt with the standard weapon. (Hey, you'll always need some bolter marines...)
Pros and Cons
The Single List
-Less effort involved in post-mortems; only ask 'how could I have done better?' instead of potentially getting distracted with 'what could I bring?' Analysis is simplified, basically.
-Easier to pack/carry
-fewer models to buy
-Easier to gain experience with a specific set of units/armies/scenarios; mastery of the list comes faster.
The Evolving List/Multiple lists
-Better idea of what any given unit of your codex can do.
-Opponents always on their toes if they're not sure what you bring
-Variety is its own reward; you can avoid being stuck in a rut
The Single List
-entirely possible to get 'stuck in a rut'; sometimes you just want a break from The List
-opponents may start getting an idea how a well-tuned version of your list works; this con's magnitude depends on whether or not they start gunning for you, as it were.
The evolving list/multiple lists
-Harder to analyze/attain mastery of the lists; you may be asking yourself 'how could I make a better list' in addition to 'how do I use my assets better?'
-More money/time involved in modeling; and you will always have some excess.
-More special rules to remember, especially if your different lists are across codices
This is one of those 'different strokes for different folks' kind of deals; I don't think there's a right answer. However, I'm more a fan of the multiple-list approach. The BT and Tau are pretty much static for me. However, between the two of them, I can radically switch the type of army I have from pure shooty to semi-shooty/punch, from hybrid to foot.
Why is this a good thing, if it makes it harder to master a specific list?
The answer is that it lets me take a break from the list. The more you play the same list, the more you adopt the mindset involved. In a sense, this is useful; you start remembering the idiosyncracies of units (IE: Eldar psychic buffs at the beginning of the turn, weapon capabilities for the Tau, all those special rules for Marines...) and making use of them.
On the other hand, it can potentially give you blinders: you start looking at everything from one POV. If you expand your point of view and know the mindset of your own army AND the army across the board from you, then you have an advantage. Knowledge is a source of power, after all.
Then, your real question is whether you want multiple lists out of multiple codices, or you want to get several lists out of one codex. Personally, I think that multiple lists out of one codex is a solid idea: if most/all of the units out of a codex are useful, then people will buy them, collect them, and get attached to the army (and thus the game). That's more of a personal opinion, though; I don't have any research to back that up.
*Any incoherency in this article may be due to the fact that I had a drink or two while blogging to celebrate A) being off for the holidays for a bit, and B) having officially picked up my MBA. HUZZAH! also, C), I have the Guinness and Irish Creme necessary for Irish Car Bombs.