*snip*
This post menaces with spikes of Awesome.
Also, I think Miuramir is the closest to correct - DF would have been considered easier to get into in 1987 than it is today because PC gamers in the late 80's were, simply put, more hardcore. The modern PC gamer is little better than a console gamer, and PC games are intentionally watered down to make cross-platforming easy. Hell, many modern PC games have the XBox 360 controller built-in with native support. I've even seen games displaying 360 controller buttons (keyboard shortcut in parenthesis) by default.
But look at any of the AD&D PC games for DOS (which, funnily enough, were ported from PC to NES frequently) - if you don't have an instruction manual, you don't know WTF you're doing. Excessively complex games have always been around - and some of them have been wildly popular. As a more modern example, take the Shin Megami Tensei RPG series by Atlus. If you are coming from Final Fantasy as a background these games are ridiculously complicated, with convoluted stories.
In 1987, DF would have been well received and featured in several magazines, but probably criticized for its lack of moral structure. In 1987, there WERE no 'ambiguous alignment' games. Those few that let you take on an adversarial role were very clear in expressing that you were bad. DF would probably receive a lot of criticism for how depraved some of the possibilities are. As Miuramir said, Roguelikes would take a line from DF, and procedure world generation for randomized gameplay would be more prominent in multiple genres, most notably strategy and RPG games, and probably shooters as well.
The medium itself however would demand that DF be a 'finished product' - that is to say, once it was released, that's IT. On to the next project. Because of this, DF itself would not have reached its 3D stage and would, in fact, be the 2D digger that set the stage for Boatmurdered and other classics.
If 2DF did extremely well, Toady might have received a contract to write StA3: Dwarf Fortress II, or have been offered several lucrative positions as a programmer for a game designer.
Projecting this into the world of today, assuming 2DF is a hit...
There would be a much stronger focus on replayability in video games. Platformers and puzzle games were the first pioneers, and both of them use static levels that can be 'mastered' with time. Ever seen a Mario 1 speed run? It's actually ridiculously short. Linear games would by necessity have to tell a better story or be longer to reasonably expect to charge the same amount as an infinitely replayable game.
There would very likely be an entire genre of 'build' games a la Sim City. Sandboxes, early iterations of build & deploy (C&C/Warcraft/Starcraft), and actual management sims putting you in charge of settlements, with or without a goal.
Procedural generation as a technology would be advanced compared to what it is now. While we know it is possible to create a game which is for all intents and purposes a Final Fantasy title but constructed out of the aether every time you hit New Game with different characters, locations, motives, goals, etc. it is just not FEASIBLE. It would take more man hours to make a competitive RPG utilizing that system and debug it than is profitable; however, if this is the basis from which most RPGs spring up, rather than Dragon Warrior, such a game might well be on the shelves of Wal Mart today.