Break the world up into sections and decide upon a data format that describes how they relate to one another; ditto for how NPC characters behave in relation to the player's progress; and so on for each element, like the savegame format and so on. Then you write a very small, two-location proof of concept quest in which each kind of object gets to exhibit the behaviour you want.
Then you turn your computer off, buy the biggest, fattest notepad you can find and design a fuller game around the technology that you now know you're capable of delivering.
For your proof of concept I would test out everything you want separately first - ie write a program that allows the player to travel from one map to another when he touches a portal, freeing the old map and loading the new one as required; have another in which to test out enemy AI and how well combat works; another to test out how to make NPC's forward the plot when the player achieves certain things, making the triggers easy so you can concentrate on the data management. You need to become comfy with every aspect separately before attempting to craft them together... otherwise you'll get a spaghetti mess.
It's quite a lot to chew on, what you're proposing. One thing you must consider is the effect of art - if you don't have access to any then working on a project like this can get very demoralising. Make sure you're prepared in terms of technical know-how and assets before diving in.
Then you turn your computer off, buy the biggest, fattest notepad you can find and design a fuller game around the technology that you now know you're capable of delivering.
For your proof of concept I would test out everything you want separately first - ie write a program that allows the player to travel from one map to another when he touches a portal, freeing the old map and loading the new one as required; have another in which to test out enemy AI and how well combat works; another to test out how to make NPC's forward the plot when the player achieves certain things, making the triggers easy so you can concentrate on the data management. You need to become comfy with every aspect separately before attempting to craft them together... otherwise you'll get a spaghetti mess.
It's quite a lot to chew on, what you're proposing. One thing you must consider is the effect of art - if you don't have access to any then working on a project like this can get very demoralising. Make sure you're prepared in terms of technical know-how and assets before diving in.