The voices in my head have decided that I am going to write a roguelike game. They are unlikely to get their way because (a) I don't have time, and (b) I am not sufficiently familiar with the space of existing roguelikes to do something genuinely new. I've played a lot of Nethack, but never ascended; I tried (Z)Angband a couple times but I was severely put off by the non-persistent dungeons and gave up on it. I've never even looked at ADOM, and I'm sure there are others.
However, I'm going to scribble down the ideas banging around in my head anyway. Maybe something will come of it. In this post I'm going to concentrate on the virtues and flaws of Nethack, which I know very well. Virtues:
- The game starts off a challenge and remains that way as far as I have ever gotten (killed by a cockatrice after finding the vibrating square but before fighting Rodney).
- The dungeon persists; that is, once you have seen a level, it will be the same for the entire game. This is self-evidently the Right Thing and I do not understand how anyone can stand to play a roguelike that doesn't do it.
- Relatedly, each level of the dungeon fits on the standard 80x25 terminal. I am of two minds about this one; there are places where one would really like to have bigger levels, notably the Quest. However, it helps the player keep everything straight in their mind.
- It is said that The DevTeam Think Of Everything; that is, just about any action you can think of works and has logical, often useful, consequences.
- There is a huge diversity of enemies, and the enemy AI is pretty darn good, although not as good as it could be.
- The game's interaction with you is witty and culturally rich.
Not so much virtues:
- The basic game mechanics are borrowed from first edition D&D and are really showing their age. The state of the art in tabletop hack-and-slash is much nicer to work with, not to mention easier to explain to players.
- I lean toward thinking that the class system is also a relic of the past that should go. I like the class specific quests but I don't like the huge variation in starting abilities (although you could argue that this is handy for replay value).
- It is too damn hard to identify items safely. If I had a dime for every time I've been killed while lugging around objects that could have saved my bacon if only I'd known what they were...
- The equipment optimization space has only a few, obvious global minima.
- Artifacts are neither as diverse nor as interesting as they ought to be. Here I have a specific idea: I'd like to borrow the way magic weapons work in Earthdawn.¹ You find a magic item, it does something not terribly exciting, but you can discover and complete side quests that unlock more and more of its powers. Naturally, the side quests get harder as the rewards get better.
- The middle game can get very tedious; lots of trekking back and forth to item caches and altars, exploring dozens of maze levels, etc.
- The overall goal is kind of boring — when you get down to it, it's just a MacGuffin hunt. If I were actually going to write a roguelike this is the big thing I would like to change.
- Starting players are too much at the mercy of the RNG for equipment. Given that you are supposed to have been the chosen one of the god and prepared from youth for the quest, and that it is known that the dungeon is dangerous, you would think your instructors would have chipped in for decent armour, or at least enough food rations that you wouldn't starve to death in the first few levels. (This interacts with the class system badly, too. If you were some punk kid with no particular skills and no holy mission it would make more sense that you have only the gear you do.)
- The code is a crawling horror, and the DevTeam's secrecy strictures make it impossible to tell whether there is any ongoing development. I infer from the way the known-bug list on nethack.org keeps growing that there is some life there, but given its size, a bug-fix release is years overdue.
¹ Which I have never played.