There's been a good bit of information distributed on spellcasting in D&D Next, most recently this; the broader ENWorld aggregation of information on wizards and spellcasting is found here. These pieces of information might or might not contradict one another, depending on how you want to interpret them. Today I'd like to talk about material components in spellcasting, starting with a little history on the topic. This may broaden into a discussion of spell design as a whole.
In 2e (and presumably earlier editions), material components for spells tended toward the pseudo-thematic, whether it was eating a live spider to cast spider climb or using glass or amber and a bit of fur to cast lightning bolt. These components had no stated price, weight, or anything else, and the components of one particular spell were never going to show up in any other spell. I don't think it ever occurred to anyone at the time to have spell components that saw use in more than one spell. Obviously, I might be wrong about this; I haven't done an exhaustive review of the 2e Player's Handbook to be sure. As more material came out for that edition, the components required for a spell were more tenuously or jokingly connected to the effect they were producing.
In the 2e games I ran - and, I honestly believe, the 2e games other people were running - we didn't track spell components. Having said that, I am now guaranteed to see at least one OSR blogger explaining that the game is only balanced and fun when the wizard spends time gathering and tracking components for each and every spell. Let's take it as read that this might work for some groups, but for people who want to get to the fun and/or plot, a playstyle in which you don't track components that do not have a price or weight is also valid, and (I propose without hard data) significantly predominant. I do clearly recall the example character sheet, though - an elven mage/thief that tracked each individual spell component (and didn't have enough spell components to make it through a whole encounter, much less a whole adventure). I don't recall whether any significant number of these spells had a costly material component.
In 3.x, WotC moved to a kind of hybrid approach to spell components. Many spells still have material components that are nods to tradition; the main point of these from a gameplay standpoint, as far as I am aware, is that the caster can be denied all of his spells with an M component by taking away all of his stuff. Correspondingly, the Eschew Materials metamagic feat avoids this limitation, and that is the entirety of its benefit. There are also a significant number of spells with a costly Focus component or Material component, such as 500 gp of diamond dust or the like. Here, the monetary cost is clearly intended to be a drawback to the use of an otherwise quite powerful spell.
Don't get me wrong, I can appreciate money sinks as much as the next MMO designer. It's just that the way this plays at the table has never seemed like fun to me. Part of that is in how the costly components are expressed: as a simple gold piece value that the game wants you to spend. What's the logical approach to roleplaying the encounter with a merchant when you want to buy 500 gold pieces worth of diamond dust? I don't actually care about the level of simulationism that haggling over that expenditure entails; I'm just trying to explain why it isn't interesting gameplay. It would have been a half-step better for the rules to say, "This requires eight ounces of diamond dust, typically priced at or near 500 gold pieces." I feel that there's still something missing, some level on which spell components are intended to represent an interesting or meaningful choice.
Which brings me to Arcana Evolved, taking another step in what I think is the right direction of making spell components represent something more noteworthy in gameplay. Specifically, many spells have optional or bonus components, such as a blue gem valued at 20 gp or higher, which grants a bonus (and often quite substantial) effect to the spell's throughput. Making the spell to be cast and the component expenditure separate choices does help a bit. Now, once you've bought blue gems worth 20 gp or higher, those are only usable on that specific spell amplification, though that amplification can be applied to various spells. In actual play, what happens is that you buy however many you feel like buying, and then treat those as a consumable pool with which to juice up your spells. They are, after all, items that have no other application within the rules than to serve as currency (but you went out of your way to buy them, so you don't care about that) or components. There are also more specific spell catalysts (let's ignore how grossly this misuses the word catalyst) that are extremely expensive ways to very slightly improve a spell. (When we've captured these catalysts from opponents in the past, we've looked at them as cash, not something useful.) Now, AE does also have costly regular components and focuses, I believe.
4e did a little of Column A and a little of Column B as well. Many utility or miscellaneous effects that previous editions lumped in with other spells were split off to be rituals in 4e, which was an interesting move. I've talked about rituals and the problems they present before, particularly in that they suggest so little flavor or atmosphere to the players that they feel like vending machines. The component system for rituals only made this worse - disenchanting items gave you a gold piece-based quantity residuum, which you could use as a universal ritual component (including making other magic items). You could go out and purchase components more tailored to the kind of ritual you were performing, but there was very little apparent reason to pursue this course if residuum was equally available. Once again, there's no real grist for an encounter with a merchant here, because if 500 gp worth of residuum costs something other than 500 gp, you have failed the most basic question of algebra and need to try again.
4e also has optional, juice-up-this-spell kind of component. Because of odd interaction between 4e's sharply scaling economy and its power system, each component includes a listing for the highest level of power it can affect. I didn't spend a lot of time even figuring out if these were worth using; there were enough fiddly decisions and if/then checks to resolve within a single spell that remembering to use a component was just not happening.
To sum up all of the above, what I don't like in the spell component systems I've seen so far is:
They seldom represent interesting choices or tradeoffs, because the item in question usually does not have any application other than "spell component;" sometimes that is narrowed to "...for this specific spell."
Acquiring the components is not an interesting interaction, in that negligible-cost components are hassle without payoff to track, and costly components are not of negotiable value. If you tell a merchant you have 500 gp to spend on diamond dust, as long as the amount of dust he sells you is actually diamond, it is intrinsically 500 gp worth; if he cheated you and your spell turns out not to work, your DM is just being a dick.
The implication of story connecting the component to the produced effect could be much stronger; even better if it somehow plugs into a meaningful cosmology. This is awfully difficult in D&D, though, as a rules set that is intended to be adaptable to any number of different setting cosmologies.
I'm going to beat the dead horse of Item 1 just a little more, because it circles back to my old idea of items in a tabletop game possessing a sense of objective reality. Does a thing feel real to the players, like something they can trade around, inspect, or use in multiple ways? Or does it feel like a checkbox on a character sheet, with no more meaning than set dressing?
My preference, then, is that spells should have optional components, and that these items are not fluid, large-count items like coinage, or things described in coinage value. As far as that goes, I think the game is best-served to give players a distancing layer of fiction between pocketbook and final effect. Let's say you want to juice up a cure serious wounds effect, because you're in a hurry and need this thing to work. Well, how about using a scroll or potion of some lesser healing effect, such as cure light wounds? You want to empower a fireball - how about using a piece of fire coral, which you could otherwise be using to enchant a weapon with a fire effect?
The obvious problem here is if the component item is available enough and the throughput effect is good enough, you'd never bother to use the component item for its original purpose, so it's not a choice anymore and it's kind of uncool again. This is the point at which I can take a page from Dust to Dust, which of course I love doing because I'm sort of insufferable about how much I like our game. (Sorry.) Ritualism makes the spell preparation step one of the most mentally intensive parts of gameplay, where in D&D spell preparation is a few hand-waved minutes at the beginning of a game-day. To make D&D operate more like DtD, shift the decision about whether or not to burn those optional spell components to the point at which you initially prepare the spell. The game fiction indicates that you're pre-casting the spells, importuning the gods, or whatever else is appropriate to your class. Maybe these rituals also include choice points like spending a healing potion to add some other benefit to one casting or all castings of a particular healing spell in a day. This creates a situation in which you will probably use the extra benefit imparted by your optional component, but you might not, and your spell slots don't carry over from day to day the way a consumable item does. As far as I'm concerned, this meets the standard of a more meaningful choice, while also using uncommon but not vanishingly rare useful items.
It's pretty clear that 5e isn't going to do what I'm talking about here, and that's fine. I do hope that their approach to components focuses on asking players to make interesting tradeoffs and on implying or outright describing a nuanced, loophole-rich cosmology. I see every reason to believe that I will want to play around with homebrewed hacks of the 5e magic system once it's published; I certainly did so often enough in 3.x, and would have done so in 4e if I had had the time and drive to invent a few more internally-consistent underpinnings.