Ultimately, it's because that kind of emergent mechanical and physical behavior is difficult for computers to do. It creates, very suddenly, a massive number of things that the game engine has to keep track of and it's only been pretty recent that computers have started being able to handle it on large-scale (such as in games like Space Engineers). A lot of physics needs to be kept track of, and more significantly it needs to be kept track of with a very high precision or else the things players build won't work reliably.moosemaimer said:Games seem to be big on physics-based construction sandboxing right now, so why not have an interface that lets you literally assemble new things out of old ones, Ikea-simulator style? Telephone handset plus lightbulb via wood glue might not be useful, but how else would you know unless you tried it? Would certainly be more entertaining and engaging than cranking out ten copies of "Ingredient H" plus "Ingredient D x2."
Look at how Kerbal Space Program lags with large rockets, for instance. That lag isn't because of the amount of texture on the larger rockets but because of the large number of physics objects that need to be tracked and more importantly how they all interact with each other.
We're getting there, it's just going to take a while for the computers to catch up.
OT: I think the real reason is that it makes for an easy way to pad gameplay. Functionally, a lot of crafting mechanics are basically fetch quests (you go out, get the things, bring them back and get a reward), but no AAA developer would be caught dead padding out a game with fetch quests. So, say it's actually a "crafting mechanic" where it works exactly the same but creates a bit of false agency so the player can feel like it's not just a device to artificially create gameplay content.