Crawl's midgame is too long!

[GermanJoey]

I've been playing Dungeon Crawl for, well, since I started grad school - so almost five years now. I've beat it every possible way to Sunday, from power-gamed Ziggurat masters to roleplaying (gigglingly) an Ogre Healer. And, for the last couple of years, my thought has been that if there's a single flaw in Crawl's game design, its that the game is just too long. Now don't get me wrong, I realize that the other big rougelikes are even longer - ADOM games can take FOREVER, and I still enjoyed playing it. (I think the time I did an ultra-ending took, off-and-on, about two weeks to finish!) The problem with Crawl's length, for me, is that it affects the game's pacing, and that's a mark against the design of such a finely tuned game as this is.

First of all, I think the early game, up until the lair and perhaps its first few floors, is just about perfect. It is challenging but fair, often very non-linear, and designed just-right to push you through a few mad-dash scrambles to survive. However, at this point, most characters are just about ready to move past a big hurdle in their development, such as a crusader that's finally pumped his Charms to have haste castable on demand, or an IE that has just gotten Freezing Cloud and Bolt of Cold, or a Stalker that has Petrify at adequate power, or Vehumet/Sif has just gifted his first book, or a Chaos Knight that has pumped invocations high enough to have Greater Servant at Excellent, or Okawaru has just gifted his first piece of equipment, or a Berserker finally has enough piety for high-powered Brothers In Arms, or a Transmuter that has ice form reliable without much Ice Magic investment, etc, etc, etc, need I go on. It is a general trend for most types of characters (except for all types of Ogres, bless their souls), that they'll hit a power plateau at some point in mid-lair. After that, the rest of mid-game is a slow, linear process of resource accumulation. For characters that can do Lair, Orc is just for gold; for characters that can do Orc and have rPois, Hive is easy food and easy items; for characters that can do Hive, Snake and Swamp 1-4 are easy; and for characters that can do these, and have, say Rmsl, Elf 1-4 is just as easy; and if a character can do Elf, they can do Vault 1-7; and if a Vault 1-7, then they can finish the dungeon. If a character dies in midgame, it is generally because they were lulled into a sense of security, and then met some nasty unique, or ran into some kind of crazy vault, or got bored and decided to rush a branch-end needlessly. It is accepted that general mid-game floor is a process of simply trudging through, collecting items and collecting XP, at very little risk. This is opposite of early-game pacing, and against what makes Crawl such a great game. And partly because of this safe but boring buildup, and partly because of all the great additions and balance changes over the last few years, the end-branches thus become much easier and safer as well. Characters that I reach, say, Vault 8 with these days, feel much stronger than how they did a few years ago.

I thus propose shortening the midgame by quite a bit:

• Lair could go from 8 floors to 5
• Snake/Swamp/Shoals/Slime/Crypt could go to 3 floors
• Elf and Orc could be lowered to 3 floors
• Vaults lowered to 4 floors
• The Dungeon, lowered to 21 floors, cutting 6 floors only from floors 14-27
• Hive changed to a guaranteed 1-floor portal

Now, while this sounds aggressive, I also propose that to counter the loss of XP and items, the incidence of Shops/Floating Vaults/Portals/OOD monsters/Uniques should be scaled up accordingly. So, for instance, if Elf were shortened from 5 to 3 floors, the chance a floor has a shop would be increased by 5/3. Overall, you'd end up with a much more interesting dungeon, the mid-game would be shortened by 3+4+1+2+1+3+6 = 20 floors (which could mean as much as an hour to an hour and a half shaved off the total game-time of a 3 rune game), and, most importantly, the game would have much more consistent pacing - therefore be much more exciting!

[TGW]

It was suggested a while ago that the size of main dungeon levels should gradually decrease as you go deeper. I think your cuts are a bit aggressive, though. Crypt to 3 and Lair sub-branches to 4 might be alright. Vaults is kind of long but there's a new layout and some monster list differentiation planned, so I'd like to see how that pans out.

[MrMisterMonkey]

early levels feel a bit big too, for how little they contain

as for the proposed cuts, I agree with them, but from my understanding, 27 is sacred or something, so floor size reduction is probably better

[GermanJoey]

I know that "27" is sacred or whatever, which is really too bad because the 2nd half of the dungeon is the blandest part of all!

[dpeg]

GermanJoey: Your analysis is spot on, but you'll have noticed that we did address the issue. Even if in a much more slow way than you suggest: we cut levels from Lair and from Hive. We plan to continue cutting.
There are some issues with extremely short branches: currently, there is a feeling of progression within a branch, i.e. you notice monsters becoming harder as you go deeper (good examples are Snake, Orc etc.). With three levels, that may be lost. Note that keeping branch depth but reducing level size would have a very similar effect. (And this is needed for Dungeon: no way we can't sacrifice the holy cow with 27 legs.)
What I don't understand is why you cant to provide xp/loot to make up the level cut. In the two cuts so far, we explicitly did not do this. Rather, the idea was to make the midgame harder by taking some xp/loot out of it. (We did make sure that the expected value of shops stays the same, however.)

[MrMisterMonkey]

I actually don't really notice monsters becoming harder in Snake until the ending vault (it could stand to be more dramatic and shortening to 3 levels would help sharpen this), and in Orc only barely so (usually difficulty is spiky, given everything under knights pretty much being popcorn, knights being decent, and everything over that being kind of rare but determining the difficulty of the branch whenever and wherever they generate -- roka band, stone giants, and warlords with xbows are particularly nasty)

Likewise, with the dungeon, along with the smallening, making lower levels more difficult (same number of monsters but more compact, more extravagant vault placement, less popcorn, more threats) should make them far more outstanding than currently (currently they are not very outstanding at all).

[evilmike]

I am always glad to see a branch get shortened here or there, and it's good that the trend is continuing (elf got 2 levels cut a little while ago, and no one shed any tears over that). Reducing level size is also a good idea... crawl's levels are bigger than they need to be a lot of the time.

What I'd especially like to see, more than anything else, is cuts to late dungeon. Basically, everything below vaults is just monotonous and dull. Vaults can be shortened to 5 levels, and everything deeper than d:15 or so could probably shrink. A couple of deeper levels could be big if they have encompass vaults, because those are always cool.

I also agree with dpeg about the xp issue... the game already has enough xp, and if you find yourself wanting for more, maybe you will be encouraged to jump into an optional branch or two (which players often do at level 27 right now). That's the case with most species, at least.

I actually like all of germanjoey's suggestions, but here is something I also think would work, and is a bit more conservative.

Slime -> 5 levels (it has 6 for some reason)
Swamps -> smaller levels (they are big and boring, making them smaller would help... the only interesting level is swamp 5).
Late dungeon (especially post-vaults) -> same number of levels, but smaller as you go deeper
Hive -> make it a portal vault, or make it 1 level like the hall of blades is
Vaults -> smaller levels as you go deeper
Lair -> again, smaller levels as you go deeper

More drastic would be to cut lair and vault down to 5 levels. This would also make most branches 5 levels long, which is a nice symmetry.

[MrMisterMonkey]

I actually think Vaults levels should start small and get bigger with more ranged threats in open spaces. Cutting it down a few levels might help if it's still dull, too.
Contrast with Dungeon levels, which should get smaller with more melee threats in tight spaces.

If Hive was a one-level branch, people would still save it for when they needed food or piety and bees are boring pushovers (not to mention the branch's monotony); timing it or putting it inside of a dungeon level would help a bit, but I think the best way is to spread it out in the form of bee rooms, such as not to be a dreadful monolith of monotony.

[GermanJoey]

GermanJoey: Your analysis is spot on, but you'll have noticed that we did address the issue. Even if in a much more slow way than you suggest: we cut levels from Lair and from Hive. We plan to continue cutting.
There are some issues with extremely short branches: currently, there is a feeling of progression within a branch, i.e. you notice monsters becoming harder as you go deeper (good examples are Snake, Orc etc.). With three levels, that may be lost. Note that keeping branch depth but reducing level size would have a very similar effect. (And this is needed for Dungeon: no way we can't sacrifice the holy cow with 27 legs.)
What I don't understand is why you cant to provide xp/loot to make up the level cut. In the two cuts so far, we explicitly did not do this. Rather, the idea was to make the midgame harder by taking some xp/loot out of it. (We did make sure that the expected value of shops stays the same, however.)

First of all, I apologize for not mentioning that you have taken some steps to addressing the issue. However, I felt that was just too small to make a noticeable difference; the lair cuts only delay the safe midgame by a little bit, and the levels cut from the hive only deprived the player from a few hundred XP and a few potions and scrolls.

Second, I think you misunderstand me. I agree that for the most part the lost XP and loot should not be made up, only that the dungeon becomes more... interesting. The effect of a greater density of vaults will indeed by more XP and loot than a similarly-sized set of plain levels, but keep in mind that the increased interestingness is in context of the pretty deep cuts I was suggesting.

But, as for your last point (which you actually made second), about branch difficulty progression, I just don't see it for the most part, not counting the branch-ends. I sort of see it in Lair, where true that you see death yaks and hydra more often deeper in, and the Vaults, where various giants show up more deeper in. The best example for level progression I think is actually the Tomb, where you're faced with a steady progression of five phases: the sphinxes, which smite-smite-smite but cast no torment; the rest of Tomb:1, which first pits you against a couple greater mummies and priests, but not in an overwhelming manner; the start of Tomb:2, with the ambush at the stairs, where you're faced with a similar threat level as before but now in a much more difficult to handle manner; the rest of Tomb:1, where you're once again facing foes on an individual basis, but now that they're universally greater mummies; and then finally, the onslaught of Tomb:3, which is again similar to phase 3 but much more difficult by the overwhelming numbers. However, in the rest of the branches I do not see much progression. Snake, IMHO, is the worst offender here. In Snake 1 you'll likely see black mambas, naga warriors, and naga mages, and maybe the occasional anaconda. The next 3 floors are indistinguishable from this, outside of the odd unique. It is only until the last floor of Snake where the threat-level of the monsters increases, in the form of teleport-othering guardian serpents and the fearsome, hasting greater nagas, but it is an abrupt transition. I could say the exact same thing about the Swamp, Shoals, Elf, and Crypt. Trimming from the middle floors in these branches would be, in my opinion, similar to the effect of trimming floors from Hive.

[GermanJoey]

I am always glad to see a branch get shortened here or there, and it's good that the trend is continuing (elf got 2 levels cut a little while ago, and no one shed any tears over that). Reducing level size is also a good idea... crawl's levels are bigger than they need to be a lot of the time.

What I'd especially like to see, more than anything else, is cuts to late dungeon. Basically, everything below vaults is just monotonous and dull. Vaults can be shortened to 5 levels, and everything deeper than d:15 or so could probably shrink. A couple of deeper levels could be big if they have encompass vaults, because those are always cool.

I also agree with dpeg about the xp issue... the game already has enough xp, and if you find yourself wanting for more, maybe you will be encouraged to jump into an optional branch or two (which players often do at level 27 right now). That's the case with most species, at least.

I actually like all of germanjoey's suggestions, but here is something I also think would work, and is a bit more conservative.

Slime -> 5 levels (it has 6 for some reason)
Swamps -> smaller levels (they are big and boring, making them smaller would help... the only interesting level is swamp 5).
Late dungeon (especially post-vaults) -> same number of levels, but smaller as you go deeper
Hive -> make it a portal vault, or make it 1 level like the hall of blades is
Vaults -> smaller levels as you go deeper
Lair -> again, smaller levels as you go deeper

More drastic would be to cut lair and vault down to 5 levels. This would also make most branches 5 levels long, which is a nice symmetry.

The one thing about reducing level size is that large levels present a unique kind of challenge - the stairs are further away on average from any given spot you might be on. Running from ranged threats now becomes much less possible! Furthermore, the vast openness of the level means that running from any kind of threat might mean you run into some other enemy from the other direction!

[MrMisterMonkey]

by keeping Vaults levels large and open (with ranged threats) and Dungeon small and compact (with melee threats) would preserve this and help differentiate these branches

note, too, that it's hard to run away when everything is clogged up with ettins and titanic slime creatures

[Stormlock]

Making levels smaller will make teleport scrolls less useful and sneaking by tough enemies much more difficult. We need to be careful we don't turn half the levels in the game into arena death matches.

Also keep in mind, shortening the midgame can be crippling to characters that have trouble getting geared for the endgame. Frankly, if you feel the midgame is dragging on too long and nothing is a threat, you can just forgo totally exploring every level and branch, and just go ahead in the dungeon. Characters that feel like they aren't ready aren't going to have the opposite option if they're running around with only a couple resists and no useful wands because theycouldn't paw through half a dungeon's worth of loot.

[psyshvl]

I like Joey's proposal, even if it's a bit drastic. The mid game defiantly needs to be shortened, though it should probably be done so slowly. It's true that shortening it would hurt challange classes (Ogres), but the proposal does include keeping the same amount of loot and XP, and even if such a change were to make the midgame harder, midgame challange classes would still have an easier time than in the early game. All an increase in difficulty for the midgame would mean is that you might have to actually burn through a few consumables or go elsewhere to get XP and loot.

[KoboldLord]

While the tightening up involved with removing the last two levels of Elf and Lair was a good thing overall, I think I prefer to see the cuts being made with a scalpel than a broadaxe. I think Hive is up next on the chopping block, and appropriately so, but one level I'd like to see less of is the very first one. The monsters are mostly melee bruisers with a few dart kobolds, and there's precious little design space for anything more complicated because the player doesn't have many or any abilities of their own to work with.

[galehar]

I actually think Vaults levels should start small and get bigger with more ranged threats in open spaces.
Contrast with Dungeon levels, which should get smaller with more melee threats in tight spaces.

This is a great idea for differencing Vaults and Dungeon. What Vaults really need is its own monster set. Once done, we'll see if it also need shortening.

I think the problem with the dullness of the middle game comes from the big gap in difficulty between normal level and branch end vaults. This leads to the optimal play being "clear dungeon and branch in ascending order of difficulty, skipping branch ends and levels with nasty uniques. Then get 3 runes and go to Zot."

currently, there is a feeling of progression within a branch, i.e. you notice monsters becoming harder as you go deeper (good examples are Snake, Orc etc.)

I guess the problem is that this isn't noticeable enough. Sure, I remember fleeing from a guardian naga on Snake:4, or an orc sorcerer on Orc:4. But it's too rare. So maybe we should boost the difficulty of lower branch levels. More high level monsters and/or some nasty vaults.

[dpeg]

First of all, I apologize for not mentioning that you have taken some steps to addressing the issue. However, I felt that was just too small to make a noticeable difference; the lair cuts only delay the safe midgame by a little bit, and the levels cut from the hive only deprived the player from a few hundred XP and a few potions and scrolls.

No need to apologise. I wanted to point out that we're aware of the problem, and not in the "should be fixed some time" sense, but rather in the "we've started doing something about already".
Of course it is your good right to demand sweeping changes. But suppose for a moment that you were in the driver's seat. Would you still cut all those branches in one single step? I think it is inevitable that developers tend to be more conservative than players (this includes a strange transformation when a player becomes a developer -- I've seen this, and experienced it myself).

Second, I think you misunderstand me. I agree that for the most part the lost XP and loot should not be made up, only that the dungeon becomes more... interesting. The effect of a greater density of vaults will indeed by more XP and loot than a similarly-sized set of plain levels, but keep in mind that the increased interestingness is in context of the pretty deep cuts I was suggesting.

But, as for your last point (which you actually made second), about branch difficulty progression, I just don't see it for the most part, not counting the branch-ends.

So we need to spend thought about how to do this better. Regarding Snake, I always like it how Snake:1 generally has mostly snakes and few nagas, and as you do deeper, you see more and tougher nagas.

[dpeg]

Regarding smaller levels: some of you seem to think that we're talking about 20x20 (standard size is 80x72). That's not meant... already decreasing smoothly from 80x72 to 50x40 would have a drastic effect, without crippling the player automatically.
Smaller levels have the fun possibility that they don't need to be rectangular any more... organic levels like Swamp could be even better, with patches extending out of the main area and so on. (Same for vaults, where rooms could extend beyond the main level.)
Regarding stairs: there is no reason why we should feel forced to keep three staircases in each direction. There could be fewer staircases, and more hatches, for example.

[evktalo]

Very nice thread. I agree with midgame being too long, and that the optimal strategy of sparing the branch ends until you've grinded enough regular levels should be gotten rid of. Cutting floors without compensation feels like a good idea to me - you'll eventually need to go to a challenging place. More smooth difficulty progression from the first floor to the ending is also a good idea. I like the points dpeg makes about levels getting smaller in the last reply very much, especially the non-regularly shaped levels and effects of vaults to the level geometry.

This is a great idea for differencing Vaults and Dungeon. What Vaults really need is its own monster set. Once done, we'll see if it also need shortening.

Here's a start.

--Eino

[GermanJoey]

First of all, I apologize for not mentioning that you have taken some steps to addressing the issue. However, I felt that was just too small to make a noticeable difference; the lair cuts only delay the safe midgame by a little bit, and the levels cut from the hive only deprived the player from a few hundred XP and a few potions and scrolls.

No need to apologise. I wanted to point out that we're aware of the problem, and not in the "should be fixed some time" sense, but rather in the "we've started doing something about already".
Of course it is your good right to demand sweeping changes. But suppose for a moment that you were in the driver's seat. Would you still cut all those branches in one single step? I think it is inevitable that developers tend to be more conservative than players (this includes a strange transformation when a player becomes a developer -- I've seen this, and experienced it myself).

Second, I think you misunderstand me. I agree that for the most part the lost XP and loot should not be made up, only that the dungeon becomes more... interesting. The effect of a greater density of vaults will indeed by more XP and loot than a similarly-sized set of plain levels, but keep in mind that the increased interestingness is in context of the pretty deep cuts I was suggesting.

But, as for your last point (which you actually made second), about branch difficulty progression, I just don't see it for the most part, not counting the branch-ends.

So we need to spend thought about how to do this better. Regarding Snake, I always like it how Snake:1 generally has mostly snakes and few nagas, and as you do deeper, you see more and tougher nagas.

I do appreciate you being conservative with design changes. But, like you said, as a player, and especially one who has run through the same mid-game doldrums so many times, I tend to have more empirical bias and personal experience to make me feel more strongly about it. That's how it goes!

[Zicher]

Why is it that so many players find the midgame boring?
Is it just me, who finds the midgame (non-final Swamp, Shoals, Snake, Vaults, Elf a Dungeon itself) being a preparation to see if my char can stand the appropriate bottom-level challenge?

[GermanJoey]

Why is it that so many players find the midgame boring?
Is it just me, who finds the midgame (non-final Swamp, Shoals, Snake, Vaults, Elf a Dungeon itself) being a preparation to see if my char can stand the appropriate bottom-level challenge?

Because it typically lasts 4-6 hours out of a 8-10 hour game and challenges are rare and often startling, as you've been lulled into a sense of repetitive security. I guess whether you like the mid-game reflects how you feel about games that typically require lots of grinding.

[KoboldLord]

I don't experience boredom in the mid-game. I have my choice of several tactically distinct branches to do next, my basic playing style has come together, and I'm not too worried about being one-shotted from the edge of line-of-sight. On the other hand, I probably do not yet have the perfect set of artifact jewelry or full coverage of emergency escape abilities, so I will still die very fast if I react to a new threat incorrectly.

I sometimes experience boredom in the early game, because the outcome is occasionally genuinely outside my control. A freak wand of draining from around a corner on D2 is not a solvable puzzle, and it means that gameplay up to that point was a complete waste of time.

I also sometimes experience boredom in the late game, particularly in Pan but also sometimes in Tomb. By this point in the game, the player has all the tools they could want to enlarge their comfort zone and make their character safe regardless of what they face. Pan and sometimes Tomb don't do quite enough to push that comfort zone back to interesting levels. I have gone into detail before, so I probably won't go into detail here.

Trimming parts of the midgame to make room for a greater variety of content is appealing to me, but not because of any specific problem I have with the midgame. It's more a matter of trimming a few ounces off a serving of high-quality steak to make room for baked potatoes and steamed vegetables to be part of the meal with it.

[Grimm]

new vault idea: the Ruth's Chris vault

features cow ghosts and animated broccoli warriors

[zasvid]

I guess whether you like the mid-game reflects how you feel about games that typically require lots of grinding.

I guess it doesn't have to. I generally learned to hate grinding in games and I have yet to feel bored in Crawl's midgame [¹], after a year of playing. Though I can imagine it might happen after beating it a large number of times - I've experienced such a change of feeling playing other roguelikes. I feel it only becomes a problem for really experienced players. Of course, it's worth to address it and converting long low-risk, low-reward parts of midgame into shorter high-risk, high-reward bits sounds like a very good solution, though I am not entirely sure that it is what has been proposed in this thread - the alternative interpretation of a few statements in this thread would be amping up the difficulty of mid-game and that might change it from boring to frustrating instead of exciting.

Also, this discussion stimulates me to ponder deeply upon the problem of the midgame in roguelikes, thanks for that

[¹] On the other hand, I have been bored in the early game or rather after a string of deaths in the early game, because the lack of options there is often painfully frustrating.

[jejorda2]

I don't like the idea of making Hive a portal vault. That makes it too difficult to carry all the food out. I guess carrying a potion of levitation in there isn't too onerous.