Yeah writing my own engine was a hard decision. On one hand, I hate writing boiler-plate code, but on the other I have become so used to it at this point that I can whip up a basic rendering engine in a few weeks (the most recent iteration of my engine was written in two weeks of free time, in about 80 hours). I have wanted to use existing engines like Unity, but I have found that they pigeon-hole you into a specific type of industry-standard workflow that is not really good for me (for example I do not use any external assets, everything is generated in-engine since I am not a very good artist).

That said, I'm just kind of taking everything one step at a time. When Minecraft was first released, it was just a simple block-based world editor; I intend to release the simplest thing I can first and work from there. As for the AI, it is easier than it might seem to implement and I am not doing anything that novel (primarily the deduction is based off of languages like Prolog); I have implemented basic (test-based) prototypes in the past and I am pretty happy with the results so far.

I'll have to check out First Earth; as for multiplayer, I am prioritizing it last - not because I do not love a good multiplayer game, but because it is the one thing I can safely remove and still have a fun (enough) game while greatly reducing complexity.

Thanks for your thoughtful insights. :)