World may does not need to be transformed into cubes, it can be kept as what it looks like originally...... I mean it would be more interesting if the game mode of minecraft is intruduced
I don't think it would be fair to leave you with a "it's not possible, this is stupid" kind of response, so here is a limited answer. Take in mind that I am not in any means a specialist in real time rendering, so it's more or less a proselyte explanation.
If you want to modify elements in three dimensions, you usually need a voxel system, where the world is stored as single sized elements (cubes in Minecraft) that can be modified individually.
In Minecraft, each cube has fixed dimensions (1*1*1 meters). Each cube is stored with it's specific informations (it's type, position, orientation and state). Cubes are grouped in a chunk system, each of these being a 16*16*128 blocks element (chunks are here to simplify saving and loading of the world, it's easier to load a single bigger file than trying to determine which cube need to be loaded or saved). So it's already a certain amount of data to load, and I'm not even talking of the geometry.
Now, the tricky part. I think Minecraft loads around 16 chuncks max of terrain around the player. It's 524288 blocks, each having it's own informations stored. Now, 16 chunks represent a 256 meters distance. Just think about the amount of data that will be required to process to render something with the view distance of Outerra. You could go with a mixed system, where only the close distance terrain is based on voxels and the long distance terrain is using standard geometry. Problem is, if you give the player the ability to dig in the terrain and place blocks, you need to be sure that it's possible everywhere. In Minecraft, the world height is 128 meters. In Outerra, well, even if you set a standard as high atmosphere and start counting the earth depth, it's gigantic.
Let's imagine we have a system powerfull enough to store this amount of data and render it in real time. There is yet an other problem: as others use said, Minecraft terrain is flat. Outerra is rendering earth a curved terrain, and the dataset has to be processed to be compatible with this system. As a result, you can't have a single dimension type of cube, because it will be incredibly big, up in the atmosphere (where the earth dimensions are larger) and could go to extremely small dimensions deep under the surface. It's a problem impossible to solve if you want the player to be able to manipulate the environment.
There is still the possibility of an hybrid system. Even if you keep a mostly fixed world and don't use voxels for everything, you can create a building system where the player can create a first block which will be used as the origin point of the model so it can be located in the world, then use a specific data system to determine what type of cubes are attached to this first element. I think it's the system used by the Spanish dev who made the tool used in Everquest Landmark. But in the end, with something of Outerra size, it's really difficult to say what kind of limitations had to be put on a system like this, because you will end-up with the same problem: if there are too much of these voxels to render, the game will start to lag.
I hope it helps you to understand the problem here.
Annnnnd here goes my break