Labyrinth

Labyrinths are much better than most or all of the other portal vaults in two ways: they are flavorful and they are balanced. Hunger is a well-defined threat that affects powerful and weak characters equally. But in another way, they are much worse: to most players, they are boring. Finding a sewer or bazaar is usually much more exciting than finding a labyrinth. This is mostly due to two things. One, they are very predictable; you get a maze of one-tile-wide cardinal corridors and one minotaur. Two, they are not like the rest of Crawl; they involve the player solving a maze like they normally would, and that's pretty much it.

I am not sure about the “most” but otherwise I agree. (In general, you hear the complaints the loudest.)
Note that we have very recently changed labyrinth generation: there will be at most one in the whole dungeon (chance to see one is 70%), and it will spawn only in D:10-20 (plus Elf or so). The idea is that early labyrinths are more interesting (rush, food, minotaur). This should already help a lot. Of course, labyrinths should still be improved further. — dpeg 2011-01-03 12:14

Another aspect of labyrinths that makes them tedious is that, unlike every other portal vault in the game, you can't easily leave them. You can leave a sewer if you think you can't clear it, and you can leave a sewer if you decide the sewer is boring. You don't get that with labyrinths; if you enter one, spend a few minutes and decide you're sick of the labyrinth, you can't just go back to the dungeon. You have to finish it. This is also seen with Abyss and Pan; you have to wander around until you find an exit, even if there's no way the area can kill you.
Now, obviously this is a huge part of labyrinths both gameplay and flavor-wise, since being able to leave at any time would eliminate the starvation threat. Having multiple exits would do the same but to a lesser extent. minmay 2011-01-14 19:28

After discussing a bit on the forum (tavern thread 4408), here is a list of suggestions by dpeg, ranging from modest to more elaborate:

1. Map changing should not depend on rotted map section anymore. Perhaps different message if nearby.
2. In addition to the level changes and sounds, walls can turn translucent. In order to make it easy on console to distinguish wall types (they all look the same once transparent), use small windows (and perhaps many of them) rather than large windows. The goals of this are two-fold: it is an easy and cool visual effect, and it makes the search easier.
3. Distinguish between floor types:
   • current colour/tile: never stepped on
   • different colour/tile: stepped on recently
   • still other colour/tile: stepped on some time ago
4. Each level sound has a chance to wake up the minotaur anyway. Level sounds if he does.
5. New monster: swift, will flee to the minotaur once it sees you (and wake the minotaur up).
6. If the minotaur is awake before you arrive, she may prepare an ambush. The minotaur could hide behind a corner or, easier to code, camouflage (i.e. be a wall) until you're there.
7. Most ambitious: have walls and sections of width 2 and 3 at times (but locally, not for the whole maze). This needs care about the labyrinth generation algorithm and the code which tweaks the maze.

A player can reliably solve a labyrinth by following a simple exploration algorithm, such as “hug a wall” or “always turn left.” The result is that labyrinths take nothing but time from the player, barring the occasional monster. While solving them this way is not optimal, this is only meaningful if the player is exceptionally low on food, which is unlikely. — minmay 2011-03-25 16:03

• Make the mazes less uniform. Add diagonal corridors. Add corridors at funny angles. Add smooth bends. Add large, irregular paths, like those in Lair. Add rooms with doors. Of course not all of this is easy to code…
  • It would already help a lot if there occasionally were paths of width 2 or 3. Given what our mazes look like, I bet that width 3 is a lot easier than width 2. — dpeg 2011-01-03 12:14
• Add more threats, so that you can reliably expect to need to fight quite a few monsters. From a flavor standpoint, this can simply be other adventurers lost in the labyrinth. Since labyrinths have a huge depth range, these would probably need to be adjusted depending on the depth the labyrinth was entered from. Let's not restrict this to player races; let the occasonal giant get in on the fun. This would really break up the monotony, but it presents an immediate problem: monsters leave corpses, and corpses are food. We surely don't want to get rid of or soften the hunger threat. No idea how you could get rid of the corpses in a believable way.
  • Why not just make them the spectral/zombie/skeletal version of other lost adventurers? – psyshvl 2011-01-02 03:06
    • That would be an elegant and flavorful fix, but spectral/zombie/skeleton monsters are all generic melee opponents - boring uniformity is what we're trying to get away from. – minmay 2011-01-03 01:56
      • They could be ghosts generated with the player ghost code, or be a new type of undead (Lost soul? Wandering revenant?) that only spawns in Labyrinths, and has some unique abilities, spells, or gimmicks to make them stand out. – psyshvl 2011-01-03 02:17
        • Monsters should have gimmicks related to the labyrinth environment and its threats: Disorientation and (relating to that) starvation. Mimics that pretend be doors and labyrinth exits, hungry ghosts and any sort of trickery with the player's sense of direction (for. ex. rotating the map 90, 180 or 270 degrees around the player) might be good ideas. — coolio 2011-01-03 11:18
          • Interface screws are not good ideas. — mrmistermonkey 2011-01-03 11:19
            • Just to be clear, I didn't mean there'd be effects to the player's movement (i.e. pressing up wouldn't cause the player to move left) in any way, only on the map. Would any monster-caused effect that causes the map to change (more or less drastically) be considered bad? — coolio 2011-01-03 11:41
              • Superficial map changes such as rotating are bad. Labs already shift and deform out of sight, so perhaps simply amplify that. — mrmistermonkey 2011-01-03 12:08
  • We already have trap spiders and hungry ghosts. I would like to see more of those, as well as harpies (via vaults), sphinxes (for deeper labyrinths), eyes. There could be a Theseus unique. — dpeg 2011-01-03 12:14
  • When in doubt, why not turn to magic? Have anything that dies in the labyrinth leave a rotten corpse, with some canned message like “The oppressive air of the labrynth speeds decomposition.” That solution has some fringe effects; ghouls entering labrynths becomes extremely risky but possible for the truly brave (a plus in my book), and the saprovore mutation (which I've always found a bit underwhelming, especially compared to carnivore) makes the food problem less severe, but for the average player nothing is gained. With that I'd agree with everything dpeg said, especially normal eyeballs (wouldn't that be a fun way to starve to death?), hungry ghosts (for flavor at least, as they are right now they're not very dangerous) and harpies (who will constitute a massive threat since they attack food in inventory, thus hurting the most prepared players the most. You'd want porridge whenever possible). I'm tempted to suggest level-appropriate mummies as well; they move slower than you so you can lose them in the maze, but if you go the wrong way you may have no choice but to kill one and suffer its curse. — brickman 2011-01-04 01:12
• Remove the teleport traps in vaults. If teleport traps are really desired, make them randomly generated throughout the labyrinth (if this is really hard to code for some reason just add a minivault which is nothing but a teleport trap). Currently, these vaults are horribly spoiler-reliant. — minmay 2011-01-13 01:40
  • You really don't see any difference between random path-breaking teleportation traps and great one-time gotcha! vaults? There's no sense of “spoilers” here anyway, the things exist to be discovered. — og17 2011-01-13 05:31
    • I see a difference, which is why I think it should be changed. I completely understand the schadenfreude purpose of the vaults. That doesn't mean I like it. — minmay 2011-01-14 19:28
• Another possibility is to simply greatly increase the number of vaults, including more with actual threats. Although this might actually weaken flavor since the labyrinth is thought of as a huge and nearly featureless expanse, it would certainly make things more interesting. We'd need to actually make more vaults, obviously, but I guess I can do some of that if this goes through. — minmay 2011-01-13 01:40

I don't get this push to add a bunch of monsters to labs when the very thing that makes the portals unique is that there's so little focus on killing monsters before the minotaur. I don't find labs boring, at that - their only problem is that they're so much more common than the other types of portals. So while the fourth lab in a single game may get tedious (which is already addressed), the answer to that isn't “make them like every other area, also with a maze.” Labs have a great atmosphere as they are, there's no need to replace that minimalism with clutter. — og17 2011-01-13 05:37

If they were just random enemies that'd be one thing, but the enemies are being chosen to exacerbate the one threat the labrynth has–starvation. And I disagree with you; as it is I have never once come even close to starving in a lab, ever, just by always keeping a ration or two in my inventory when I leave my stash plus whatever I find as I go. Labrynths are fairly boring for the prepared. Harpies, eyeballs and hungry ghosts would be exciting additions, because they're not attacking your health–they're attacking your food. — brickman 2011-01-13 20:46

By vladimirdx on Tavern, posted by dpeg 2011-01-14 12:28: Labyrinths: I like them. I also strongly dislike maprot, and it has no place in the game. So how can you make mappable labyrinths non-trivial with autoexplore, requiring though to navigate? The race to the portal is fun. So make the Labyrinth also a race. You can't autoexplore or dawdle. You want to get to the Minotaur as soon as possible. Some ideas:

1. After a certain amount of time, you're get teleported OUT, with warnings along the way. No awesome loot for you.
2. The Labyrinth causes you to hunger much more quickly, making it a resource drain.
3. if you stay too long, you can get bad mutations such as fast metabolism. (I don't like this one)
4. your other stats (HP, STR/DEX/INT, maybe EXP) also drain while you're in the Labyrinth. Maybe, which one drains depends on the specific labyrinth generated. This also leads to a resource drain. Meh.
5. When you enter, you immediately become hungry (or very hungry) and can't increase your hunger level above that. You're always at the brink of starvation. (or: you can't eat in the Labyrinth. I don't like this as much.)
6. At the exit, there are 3 piles of loot. If you get there very quickly, all three are still there. If you were less quick, one or two will be there. If you were really slow, there won't be anything left.

My vision of the labyrinth is something built around hunger, around rushing, around the food clock, or imitating that design pattern. Doing so would make the Labyrinth's unique challenge work without diluting its special flavor. Minivaults and monsters are also good: I can imagine a player evading, not fighting, monsters in the Labyrinth to get to quickly disappearing loot, or facing the interesting choice of whether the items in a potential vault (which should also disappear) are worth possibly losing loot at the end for. The 'outrunning monsters' part, a lack of maprot, an emphasis on getting out quickly, and the challenges from resource drain and the inability to become 'not hungry' all might appeal to people who don't like labyrinths.

While I don't support all of this, it is certainly a new twist on the mazes. The race idea is interesting, but I think that there's a fundamental clash between “use hunger as the driving matter in labyrinths” and “starvation deaths are just not cool”. Here is my proposal to solve it: if you're starving in a labyrinth for long enough (say ten or twenty turns, to give players a chance to note this and eat), you're just expelled. Don't worry about rationalisation, that's generally easy to do. This rule would allow is to ramp up the hunger rules in labyrinths: fewer nutrition from food or, what amounts to roughly the same, higher metabolism in labyrinths. (The latter is a bit better if we add giant eyes to the monster list.)
Personally, I like the actual act of maze solving, and I'd rather like to keep the mechanics of choosing fights and using advanced tactics to other (portal) vaults. It should be possible to make the labyrinth mutation more interesting. Perhaps by mutating more, and having each mutation add new holes (with some chance). That way, the labyrinth becomes easier with time. — dpeg 2011-01-14 12:49

Autoexplore is disabled in labyrinths and the map rots quickly. Is this really needed? Players can (tediously) make their own maps; they can even do so via character dumps. The justification given for maprot is that the walls deform when out of sight, but I don't think it's a big problem if the map happens to become inaccurate; boring beetles dig through mapped areas all the time, and it doesn't create any fuss. As for autoexplore, I believe the justification is that it makes the labyrinth too easy, but frankly, if the exploration algorithm can solve a labyrinth then the player can solve the labyrinth. — minmay 2011-03-23 23:41

The history was like this: ancient labyrinths (pre-Soup Crawl) were trivial to solve (they were spirals with minimal braiding); I forget if they had maprot. greensnark added the current labyrinths, with the rock-stone-metal pattern and much more braiding; this one definitely had maprot. jpeg added self-changing to labyrinths.
Of course, old versions had no autoexplore at all, so the question was moot. Perhaps the current player base is so used to autoexplore that the idea of manually navigating the labyrinth is completely outlandish to them?
I won't let labyrinths go for these reasons (they're ancient and so many folks put work in them, they are also the portal vault prototype) and for others (they add some historical flavour to game, of the classical/oriental type I like: labyrinth/Minos, ziggurat, bazaar, etc.). So we should, or at least I will, look for ways how to make them more interesting. Here's a list taking into account ideas from others above. — dpeg 2011-03-24 00:55

1. Expel starving players (after around 30 turns of starvation, say). This allows us to play with much tighter food clock.
   • I don't like making hungering desirable to labyrinth-sick players; it's affected by so many good things (which, to avoid silly gaming, should be strictly good, in my opinion), and beyond that it's gameable by carrying something that can greatly hunger, like an amulet of rage, ring of invisibility, ring of hunger, ring of negative intelligence (and casting big spells), staff of channeling, or some such. Also, mummies can't benefit from this. If you want to make escaping the labyrinth easy, perhaps provide an (a)bility or consumable or some such that induces starvation and exits. — mrmistermonkey 2011-03-24 03:51
     • You seriously lack misanthropy and schadenfreude, old man. What about multiplied hunger and no expelling? You know, to make some folks actually die… — kilobyte 2011-03-24 10:08
2. More hungry ghosts.
3. Have the minotaur come after you (he knows his maze very well, of course). Perhaps he can trigger labyrinth changes so as to have an easier path to you?
4. Labyrinth changes could have a chance to open up more ways. This way, it gets easier to longer you stay in there.
   • Also, if a player gets really sick of the labyrinth, this way they can mash 5 for a while to make things faster - suboptimal but convenient, like autoexplore. I like this. — minmay 2011-03-24 01:56
5. There could be a couple sectors in a labyrinth, each with a fitting sentinel (minotaur zombie/skeleton). If you kill that one, the area in this sector does not change anymore, and won't maprot.