In the "sheer complexity" department, imagine trying to depict any of the following well graphically: randomly generated artifacts made of random materials, the ridiculously intricate damage models, and completely randomly generated titans (which do start with a base creature, but have random materials/bodyparts galore). While you could just do essentially what he does for the ASCII and have one symbol that represents all these, he doesn't seem to want to go this route. Which is insane, but hey, insane seems to be the normal for DF development.
The second part is that his last game died partly from his trying to implement graphics too early. Notice how the full title is "Dwarf Fortress: Slaves to Armok"? Armok was the game before this. Similar in overly ambitious scope, but with an early 3d style graphics. However, for some reason or another, it became to difficult to support its development after graphics, and it collapsed under its own weight. Hence the switch to ASCII: he could continue implementing hundreds of features without as much regard for performance or implementation. It's easy to add in dozens of new creatures, random titans, genetics and randomized traits when you don't have to actually worry about representing it. I highly suggest reading his devblog (http://www.bay12games.com/dwarves/index.html) just as inspiration for what one crazy ambitious man can do.
As for why he doesn't enlist more help, make it open source, etc, the answer is even simpler: this game is his bread and butter and both he and his brother live off the donations. Yes, part of it is that he wants to maintain control of the project, but if a program was literally the sole means of supporting you and your sibling, I think you'd be wary of giving away its code too. He has said it will likely be open sourced when he dies however.