It's been almost five months since I last commented on crafting systems in any detail in this blog - but no longer! Kainenchen and I were talking about... something or other, I can't recall what, but we came around to discussing some particular instance of crafting that I felt didn't constitute a crafting system at all. Okay, let me back up: I've talked before about questing after fragments of a completed item that the players then combine into a finished treasure. This is a perfectly good form of quest, and it can even be a reasonable way for magic items to come about in the game, but I've seen people argue that this constituted some kind of crafting system. Now, this is ridiculous on its face - it's a single instance of completion, not a system - but it also displays a misunderstanding of what's enjoyable about crafting in games, in a way that I've had a hard time articulating in the past.
The fun moment in crafting is when you make the decision. When you have 2-3 different components in varying amounts, which you combine into a new item - that's almost enough. The one thing missing from that is multiple options for what that completed item might be; the other completed goods should require different configurations of components, because this creates additional choice points for the player. Let's say you have two units of iron, two units of leather, and a sapphire, which would let me make a Blade of Winter. If I also know how to make Armor of the Golden General (two units of iron, two units of gold, and one runestone) and you don't have the latter two components, then you have to think a little more: instant gratification or patience? It would be hard for players to have much more agency than this.
The simpler model, in which the GM tells the players to quest after Item A and Item B get Magic Item C, requires the GM's explicit cooperation (particularly in tabletop games) in order to work. At least in most games, the GM has absolute control over the treasure that enters the game. I don't mean this in an authoritarian way; it isn't that the GM would try to keep the components out of the players' hands. It's just that waiting for the GM to remember to give you what you need strips the feeling of agency out of the practice.
I'm not normally a serious proponent of additional randomness in tabletop games. In most cases, I think a GM's planning leads to a better game experience than reliance on random dice rolls. Dice can't give you pacing or responsiveness to player choices. What they can do is absolve the GM: of responsibility and of directed choice. There's an extended subset of arguments to be made about how the GM weights the randomized tables, but that's immaterial for crafting purposes. Once the GM has written them, they represent world physics and can be spaded out. So in this case, I think randomized access to crafting components might have a place in tabletop games.
In 3.x and 4e D&D, most of what players would want to craft is overtly magical, so let's not dick around with Iron + Wood + Coal = Longsword here. Instead, let's talk about the crafting system that the two games do have. They both fall short of my above definition not because of a lack of choice - they have plenty of choice - but because they each operate on universal rather than specific components: gold and XP in 3e, and residuum in 4e. What I'd like to see instead can be summed up in the form of World of Warcraft's Enchanting system. This would involve, for 4e, a rewrite of the Disenchant Magic Item ritual and the introduction of specific components that went into each magic item in the book. The former change would give you a few items from a list of five component types, graduated by five-level brackets. The latter is much more work, though it wouldn't be too difficult to create general guidelines that guided what it cost to make a new magic item.
I would probably design such a system to offer diminishing returns - specifically, you're probably going to spend the components of 2-3 disenchanted items to enchant one new item. This is totally normal in WoW, since the magic items you disenchant come from other skills; in the model I'm proposing, it requires some additional worldbuilding. Is there just an ever-decreasing number of magic items in the world? That might be right for some campaigns, but I don't think I like that as a general answer.
One way to go here is to take a page from Shattered Isles' Dorums: at the intersection of ley lines, magical energies accrue in ways that can be harvested. Sometimes you come across a ley line juncture that is full and ready to be harvested, and sometimes another group has beaten you to the punch. Handle this with another magical table if you like.
Anyway, that's as far as I've gotten with this idea. Comments welcome.