What is good in roguelike gameplay?

[Bear]

Let's get past the technology for a moment and focus on game design.  What makes good roguelike gameplay?  These are all matters of opinion, so I expect that no consensus will emerge; but I want to have the discussion and see the opinions.

In some games items are relatively simple.  Most items do what you expect they will and that's all they do.  In other games, items are relatively complex.  Their effects can usually be strengthened or weakened, and sometimes changed altogether, eg, by blessing, cursing, overcharging, corrupting, or reversing them.  Many items can have interesting effects on the dungeon itself (eg, by breaking rocks, digging holes, freezing moats, triggering drawbridges, etc).  Are simple items or complicated items better for gameplay?

In some games enemies are very predictable and have several highly exploitable behaviors like lining up in corridors, predictable chase paths, predictable behavior in chasing you around pillars, etc.  They have known (or at least knowable) capabilities and are dangerous in known (or at least knowable) ways.  Their speed is usually fixed in some very simple ratio to yours, so it's "countable" and you know exactly when they'll move.  In other games, enemies are complicated and often unpredictable.  Their capabilities can vary tremendously, and unpredictably, due to individual variation or just because they can pick up and use magic items from the dungeon floor.  Different types of enemies exhibit different kinds of intelligence and, frequently, different motivations.  A hungry panther who wants to eat you will behave in a very different way from a hostile wizard who wants to steal your amulet.  And enemies can be peaceful or tame or hostile, and there are (sometimes) things you can do to pacify, tame, or enrage them.  Some creatures can be beneficial to the player under some circumstances, usually involving some risk (such as foocubi and nurses in nethack, quest masters, and so on).  Is it better for gameplay to have simple creatures, or complicated ones?

Some games have a simple interface.  It may be as few as a dozen or so commands.  There may be a single "use" key (assuming the items are also simple and have exactly one use).  Other games have complex interfaces and allow most objects to be used for many different things. Such games may have literally hundreds of commands, organized in a hierarchy of menus where you go to an "inventory screen" for acess to "extended inventory commands" that didn't fit into the main menu.  Most objects can be used as at least improvised weapons.  The first way is easy to learn and the second way is more flexible.  Which way is better for gameplay?  Would the availability of really good in-game help change your opinion one way or the other?  Would playing the game day-in, day-out, until you have the complicated command set committed to muscle memory make that game a "better" game than a simple game you'd played the same amount?

Some games have a very long equipment upgrade path.  You will probably replace your entire kit at least a dozen times during the game.  Other games have a short upgrade path where you can find good gear fairly early (less than 5 replacements) and then stick with it.  Which way is better?  Does it depend on the length of the game?

Some games have a short character power curve, where a "winning" character is likely to have less than 15 times the hitpoints/damage dealing capabilities of a starting character.  Low-level characters have a chance, if they are sneaky cowards, of surviving surprisingly deep.  Other games have a  long character power curve, where a "winning" character is usually at least 100 and sometimes 1000 times as tough as a starting character.  Low-level characters, regardless of sneakiness or cowardice, get killed fast if they get out of their depth.  Which way is better?  Does it depend on length of game?  Does it depend on the length of the equipment upgrade path?

[Krice]

It depends on game's design. And for that we have game designers. They are the guys who know what to do.

[Etinarg]

I like the part of equipment assembling. Finding and selecting items to have a good kit for my PC.

[getter77]

If you can wind up with a great many varied YASD and YAVP postings in theory, generally arising from a variety of builds, styles, and interesting encounters to run into, and the game has enough of a hook and replay value to encourage them---I would reckon you to pretty much be there.

I suppose in a magic world a project would start of as a streamlined and direct one...fun in turn, while then also forking off for a more complex variant twist and then keeping both pretty well in sync dev wise as then no flavor goes untasted by anybody.

[Bear]

I like the part of equipment assembling. Finding and selecting items to have a good kit for my PC.

Does a hard ID game contribute to the fun, or get in your way?  And how about the ability to make limited modifications to items or have items behave differently in different circumstances (for example the way stuff works differently in Nethack if blessed, cursed, used while confused, etc)?  Does that contribute to the fun, or just complicate matters?

[Etinarg]

ID should be available fairly conveniently for my style of play. I avoid unidentified items if possible, so games where ID is limited give me troubles. It's fair enough if ID scrolls cost money and are not always/everywhere available as long as there are overall enough of them. Or other, risk-free means to identify items are available.

Lot's of modifications are good, also context-sensitive modifications. Nethack is obscure though, since without spoilers it's very hard to find out how things work and what side effects there are. For me it's important that game mechanics are clear.

I also like to alter items like the gems, runes and runeword ideas from Diablo II.

A few of the restrictions from RL games seem silly, though - e.g. a priest cannot read prayers from books while blind in Angband. While I agree that mages might be required to actually read the spell, not just reciting it, for priests I do not believe that. So while generally I think these limitations are good, some of them appear unreasonable to me. (In general, I think the books should only be needed for learning the spells/prayers, not each time one uses them).

The "slay" modifications are a particularly good invention I think. But only if the game has themed areas where one sort of enemy is very frequent and there is a matching slay. I do not like to switch my equipment every few minutes, and to have to carry 5 different weapons "just in case".

This are just my very personal preferences. I do neither try to say what is "good" in general, or say that games must be made this way. It's just what I like and dislike among item identification and item use.

[Bear]

ID should be available fairly conveniently for my style of play. I avoid unidentified items if possible, so games where ID is limited give me troubles. It's fair enough if ID scrolls cost money and are not always/everywhere available as long as there are overall enough of them. Or other, risk-free means to identify items are available.

I think I agree in most points. In particular I like "detect curse" and "detect magic" as specific effects available early/cheap/plentiful, to remove unreasonable risk from id-by-use and allow people to not waste ID on non-magical items.  But I like ID itself rare and/or expensive, and I like items with substantial tactical components that are likely to be risky in some circumstances even when functioning as designed (beams that reflect from walls, large fireballs in small rooms, etc). "Reasonable" risk IMO is from the items working the way they're supposed to when you try them in an inappropriate situation. "Unreasonable" risk from curses etc should be possible to minimize, or with careful play avoid completely.

Lot's of modifications are good, also context-sensitive modifications. Nethack is obscure though, since without spoilers it's very hard to find out how things work and what side effects there are. For me it's important that game mechanics are clear.

Agreed, again.  Providing a reason for a character to seek to curse his own equipment, or become confused, etc, is providing an effect that benefits only the spoiled, and is for that reason unfair.  Characters should have cause from basic game mechanics, not from spoily oracle hints, to expect the modified effects as a result of the modification.

The equipment-modification subgame is a large part of what gives nethack its depth, but that subgame can be supported with other modifiers, more reasonably/intuitively connected to the modified effects.

I also like to alter items like the gems, runes and runeword ideas from Diablo II.

I don't know anything about this.  Can you give some examples?

A few of the restrictions from RL games seem silly, though - e.g. a priest cannot read prayers from books while blind in Angband.

Remember the primary issue is game design not realism, so consider the game design objective.  They were trying to balance the spell-casting classes so they didn't have a grossly unfair advantage over all other players.  The reading restriction (amongst others) did that.  If you think it's unreasonable, you don't have to do it with exactly the same set of restrictions; but you still have to accomplish the game design objective, so your spell-casting classes don't have a grossly unfair advantage.

This are just my very personal preferences. I do neither try to say what is "good" in general, or say that games must be made this way. It's just what I like and dislike among item identification and item use.

Well, personal opinions are what this thread is about.  If anyone claims that there's a real objective best game design, s/he is wrong.

[Etinarg]

They were trying to balance the spell-casting classes so they didn't have a grossly unfair advantage over all other players.  The reading restriction (amongst others) did that.  If you think it's unreasonable, you don't have to do it with exactly the same set of restrictions; but you still have to accomplish the game design objective, so your spell-casting classes don't have a grossly unfair advantage.

Maybe something like "Your god can't see the world but through your eyes while you pray" could explain why prayers do not work when blind. I understand that such restrictions spice up games. I guess it's just a matter of taste of one likes the chosen reason for the restriction or not.

The gems and runes thing should be easy to explain.

Starting with the gems. Each gem has an effect that can be added to an item. For example, chipped rubies give +10 life (HP). Items have so called sockets for gems (and runes). If you put a chipped ruby into a socket, the item will get the +10 life bonus.

Runes work similar. A "El" rune gives +50 to hit bonus (much less effect in Diablo II than +50 to hit would have in Angband, it's a ratehr small bonus) and +1 light radius. If you put this rune in an item socket, the item will get +50 AR and +1 light radius mods.

Then there are rune words. If an item has several sockets, you can put several runes into them. If you assemble a "rune word" you get all the individual mods of the runes plus one or more "rune word" specific mods.

Not all items can have sockets. I think only weapons, armors and helms can be modified this way.

[Drag]

My point is in the bottom paragraph.

Let's get past the technology for a moment and focus on game design.  What makes good roguelike gameplay?  These are all matters of opinion, so I expect that no consensus will emerge; but I want to have the discussion and see the opinions.

In my opinion conventional Roguelikes, I'm speaking Rogue-Nethack-etc are in a dead end. It doesn't get much better than Stone Soup with tiles or <insert your own favorite here>. I even managed to make a few "oh god no roguelikes!" people play it and they enjoyed it for about 20 deaths. From a technical and conventional standpoint, the dungeon-creating algorithms are refined and at times give truly remarkable results. Even useability is at a level that is "as good as it will probably get".

In some games items are relatively simple.  Most items do what you expect they will and that's all they do.  In other games, items are relatively complex.  Their effects can usually be strengthened or weakened, and sometimes changed altogether, eg, by blessing, cursing, overcharging, corrupting, or reversing them.  Many items can have interesting effects on the dungeon itself (eg, by breaking rocks, digging holes, freezing moats, triggering drawbridges, etc).  Are simple items or complicated items better for gameplay?

Items that do simple things, like killing stuff, should be simple and do what the player expect them to do. I'm not a big fan of improving a +3 sword/blaster/whatchacallit to a +4 blaster. It's just supposed to kill stuff and "stuff" should never take a bajillions hits to die unless your weapon is completely defunct. Now changing the way you wield your sword (technique) or changing the energy settings on your blaster --> Good, immersive, believeable. Items that do complex things, like tools to change the enviroment or something to create other items or even creatures should have complex mechanisms and upgradeability behind it. In summary: Basic stuff --> simple ; Complex gameplay elements --> complex; duh.

In some games enemies are very predictable and have several highly exploitable behaviors like lining up in corridors, predictable chase paths, predictable behavior in chasing you around pillars, etc.  They have known (or at least knowable) capabilities and are dangerous in known (or at least knowable) ways.  Their speed is usually fixed in some very simple ratio to yours, so it's "countable" and you know exactly when they'll move.  In other games, enemies are complicated and often unpredictable.  Their capabilities can vary tremendously, and unpredictably, due to individual variation or just because they can pick up and use magic items from the dungeon floor.  Different types of enemies exhibit different kinds of intelligence and, frequently, different motivations.  A hungry panther who wants to eat you will behave in a very different way from a hostile wizard who wants to steal your amulet.  And enemies can be peaceful or tame or hostile, and there are (sometimes) things you can do to pacify, tame, or enrage them.  Some creatures can be beneficial to the player under some circumstances, usually involving some risk (such as foocubi and nurses in nethack, quest masters, and so on).  Is it better for gameplay to have simple creatures, or complicated ones?

Predictable (beatable, masterable) enemies are good. They may be repetitious, but most people prefer it that way. A game is fun if you can "get behind the mechanics" understand it, and exploit it to full reward. However, a slight mix of the two is nice, like an odd, proceduraly generated variant of an enemy ever so often. This does not mean that there should be just enemy npcs. Neutral and sometimes friendly NPCs are a great benefit to atmosphere and believeability of any game.

Some games have a simple interface.  It may be as few as a dozen or so commands.  There may be a single "use" key (assuming the items are also simple and have exactly one use).  Other games have complex interfaces and allow most objects to be used for many different things. Such games may have literally hundreds of commands, organized in a hierarchy of menus where you go to an "inventory screen" for acess to "extended inventory commands" that didn't fit into the main menu.  Most objects can be used as at least improvised weapons.  The first way is easy to learn and the second way is more flexible.  Which way is better for gameplay?  Would the availability of really good in-game help change your opinion one way or the other?  Would playing the game day-in, day-out, until you have the complicated command set committed to muscle memory make that game a "better" game than a simple game you'd played the same amount?

Learning how to play a game is fun up to a certain point of hurt. Menus and commands have to be context sensitive. On the other hand you should never compromise depths of a game to simplify the control scheme. It is the wrong place for such considerations in these days. An good ingame-help eases things up tremendously, especially for your target audience, smart people with varying degrees of attention span and time. Interactivity and the ability to "improvise" solutions to problems the player is facing is key to good gameplay. Enable the player to tackle a particular problem in unique and freeform ways and the player has fun because "his" solution did work or resulted in so called "epic fail" (FUN) Now using objects as improvised weapons is only a tiny, tiny portion of the true potential that lies here.

Some games have a very long equipment upgrade path.  You will probably replace your entire kit at least a dozen times during the game.  Other games have a short upgrade path where you can find good gear fairly early (less than 5 replacements) and then stick with it.  Which way is better?  Does it depend on the length of the game?

How often you switch your equipment depends on the length of the game and the amount of equipment you get. In that regard I personally dislike "equipment spam" which results in me weeding through oodles of useless crap and probably missing a few gems because I lack insider knowledge to spot them. Identification of objects is only fun if there is truly something to discover! What the hell do I care about the fifth +1 ring of whatever. What I want is something of a skill check to unlock varying degrees of usefulness out of a strange artifact. A good roll and maybe you figure out a game changing use for it --> You have something to TALK about with your friends who also play the game: "Hey guys I found this truly remarkable, rare thing that..." BAM, you as a game maker caught the attention of a player and his peers, who might continue to talk about how epic the game is and make a Let's Play... and so on. That said, equipment you find should have some uses and it should be very rewarding to experiment with it. Equipment can be very powerful/rare if there is a logical explanation for it.

Some games have a short character power curve, where a "winning" character is likely to have less than 15 times the hitpoints/damage dealing capabilities of a starting character.  Low-level characters have a chance, if they are sneaky cowards, of surviving surprisingly deep.  Other games have a  long character power curve, where a "winning" character is usually at least 100 and sometimes 1000 times as tough as a starting character.  Low-level characters, regardless of sneakiness or cowardice, get killed fast if they get out of their depth.  Which way is better?  Does it depend on length of game?  Does it depend on the length of the equipment upgrade path?

I dislike becoming a shoeless god of war completely overpowering earlier foes. Progression in a game should mean that I unlock new/clever/hilarious/efficient ways of solving problems instead of (just) getting a +1 to damage/whatever. Why can't I kill a horde of undead space miners from ceti V by overloading a power conduit in a minigame. Why do I have to repetitively fire ma blaster +5 until they are all dead or my arbitrary hitpoint number is depleted? Why can't I improve my reputation with the "Stereotypically tough mercenary company" so I can call them in as reinforcements to wipe out the alien infestation on an untouched planet full of mineable goodies? So they help me further my overarching goals of galaxy domination?

Yes you read right, overarching goals in a halfway believable world. What you save on art and graphic you have to give back somewhere. I'm sorry to say, but Dwarf Fortress certainly is not as popular as it is because of its classic roguelike part! And it still is in its infancy, barely exploring the possibilities "modern" games can never hope to explore. Now I don't think the excess of DF's procedural generation is truly needed for a good/popular neo-roguelike. There is a lot to be done by just creating a game that reacts and gives clear, concise feedback on the players actions in a set-up world with limited procedurally generated content. Like different factions each game or proceduraly generated worlds (lets step away from "dungeons"?) Elements of building up what is yours, from the humble beginnings to your own "bases/castles/starships/whatever" make for a truly interest-catching experience. An experience where the player "can't get enough".

[deeper]

Exploration is what matters most in a roguelike. Unending strangeness, deeper and deeper. Weirder items, monsters and fortresses. I don't even care about winning. I just want more world.

Level generation is good. I think angband has a beta artifact generator. How about a monster generator? A vault generator? We could really have an infinitely deep world that never gets stale.

OTOH I've gotten tired of roguelikes. I like realtime action. I'd like Angband and Kobo Deluxe to have a baby.

[Etinarg]

I think angband has a beta artifact generator. How about a monster generator?

The problem is that random monsters really puzzle the player. Artifacts, and items in general can be identified so the player know what the thing can do.

A random monster is much more difficult to to judge.

"You see a Furglabaz. No battles to death are recalled. Nothing is known about it's attack".

It might have something really evil, like a nether breath, but can be harmless and it's only attack is to dim your light source. If a game has random monsters, it needs mechanics to learn about these monsters in a fairly safe way too. A deadly mistake is permanent, and the next game will have different random monsters, so one cannot learn from a past game ...

I'd assume this is the main reason why so few games have random monsters. on the other hand, I like the idea, but it seems to be difficult to do right.

[Bear]

Exploration is what matters most in a roguelike. Unending strangeness, deeper and deeper. Weirder items, monsters and fortresses. I don't even care about winning. I just want more world.

This is something I totally agree with.  I've already decided to add a bunch of "rarities" to the game.  A "rarity" is just something that cannot be generated in every game.  I mean, for example, every game can have a few ultra-powerful "artifact-level" items, chosen from a set of dozens or scores.  But more than that, every game can have a few low-level peculiar items, and a few mid-level peculiar items, a few recurring NPC's, a few unique monsters, and a few kinds of rare creature, and ...  all of these things, each chosen from its set of however many, are rarities.  They generate interest _because_ you don't see them every game.  Their appearance puts you into new content, and new content keeps happening game after game after game. 
 

Level generation is good. I think angband has a beta artifact generator. How about a monster generator? A vault generator? We could really have an infinitely deep world that never gets stale.

A good excuse for getting rare creatures (or procedurally generated creatures) into and then out of the game is to have them be the minions or constructs of a unique NPC.  In a fantasy-themed world, if Drusilla the Bone Gatherer is loose then a randomized, procedurally generated, or just plain rare type of undead (presumed to be her creations or minions) can be found with her.  A nice thematic way to introduce a rarity - a monster type which you'll meet *many* of before you manage to defeat Drusilla, and then suddenly they're all gone and you'll get the progress rush for making a win that changes the game conditions.  I think this amps up both the reasons and the rewards for defeating uniques. 

OTOH I've gotten tired of roguelikes. I like realtime action. I'd like Angband and Kobo Deluxe to have a baby.

I'm much more a game-theory purist and don't particularly like realtime games.  But it gives me an interesting thought.  It wouldn't be hard at all to make the commands in a roguelike game take the same amount of time it took the user to type them, and add a "timeout" command that keeps happening every half-second or so even if the user stops typing.  The result would be a lot like realtime play.

[Bear]

I've been thinking lately about the depth of a game and what supports it.

Dwarf Fortress has great depth, and it's supported by arbitrary player-created maps and a whole tribe of characters.  When the fortress is attacked, there are tactical considerations about routes and getting defenders to the post and decisions to make like flooding passageways and  random things that happen to individual dwarfs and so on. Each combat is an application of simple rules about dams and fluid flows and fights and accessibility/mobility and supply and so on, and players use these consistent rules plus their ability to create a map and configuration, to make a whole that either efficiently defends itself, or fails to do so.  This is a very pure application of a Conway-type game, with actions begetting reactions and consequences while the player strives for a "good" outcome as measured on several different axes.  That is really a brilliant premise for supporting incredible depth of play, and I can't think of any real parallels in roguelike games.

Chess, Go, and Othello have a different kind of depth.  It's more epiphenomenal.  The rules about individual pieces are simple, but combinations and particular formations have subtle emergent properties.  The two applications of this kind of depth in roguelike games are the equipment subgame and monster/monster and monster/terrain interactions. 

The equipment subgame in some roguelikes has some of that type of depth, in that even though each individual item is simple, forming a strategy involves knowing what combinations of items have particular emergent properties and effects and what combinations are actually possible within the constraints of what slots each item occupies.  Modifiable equipment adds a dimension to the game, but needs to be more than just an obstacle-course to a better chunk of equipment; each modification (or decision to not modify) needs to be a valid option in the development of a strategic combination. In other words, each modification ought to involve some kind of tradeoff, where some strategies will benefit from it and some will benefit more from not doing it.

Monster/monster and monster/terrain interactions also have some of that property, in that simple (but different) movement or interaction rules per monster and some varied terrain can make for a huge variety of combinations and formations with different emergent properties. 

What else supports real depth-of-play?

[guest509]

I enjoy this discussion but have very little to add. In general i play crawl, powder and doom/aliens. Grew up on old hack builds. they were all pretty fun. All so different. So it is hard to give answers. Generally a simple interface, low confusion but freakishly deep and varied is what roguelikes can bring better than other games.

[Ancient]

Since you bumped the thread I'll add my three zorkmids.

Lining up and other predictable movement behavior is fun to exploit when game features equipment making use of these arrangements. ZapM and DoomRL feature railgun. Tight long corridors are a blessing. Creatures aware of this tactic and trying to make this harder for player seem to have "good AI". Monster acts significantly smarter but not in unbalancing way. Eight way targeting also falls in this category. In MPRDB dragons are actively trying to line up with you to have a clear path to breathe fire. In general I find omnitargeting to be less interesting mechanic. It fails to capitalize as well on grid nature of roguelikes.

About identification. Most pleasant ID is when there is not so plentiful source of easy and full identification and plenty of methods giving partial information. Wands engraving in NetHack is good example. When you engrave something and get effect "the bugs on the floor stop moving" you just narrowed wand type to two different effects. If you happen to also know price of the given item you know what it is. Bad example is ring identification in ... NetHack. You drop the ring in a sink and get a message which gives the ring class away to you but without automatically identifying it. The sacrifice a ring to ID it mechanic itself is acceptable but prompting to name is just silly. Oh, and avoid potions of death please. ZapM has one currently ... :-( Angband too but fortunately it no longer instakills.

Interface? Here I think my musings will not be overly useful. I can master almost any interface unless it is context sensitive. My particularly hated behavior is Auto-Select-Doors from ADOM. Well, there it can be turned off. In DoomRL it is simply evil. 'o' key automatically opens door when there is only one in your vicinity. However, when there is more than one it asks you for direction. Combined with fast-and-furious gameplay you sometimes press 'c' to close a door to obstruct monsters' line of sight and flee other way. Now imagine you being in a vertical corridor with two open doors to the east and to the west. Some imps are on west side. So you press 'c' and go east for supercharge globe. But! Your "move" east was interpreted as close the door to the east. Bang! Boom! You are dead. Killed by interface. Close command asking for direction is fairly rare. This inconsistency prevents you learning interface. Correct solution is to ask for direction for open and close *every freaking time*. You will get used to and this will never fail you. Your fingers will know this by muscle memory.

I've been thinking lately about the depth of a game and what supports it.
What else supports real depth-of-play?

Monster and item descriptions. They create unique world and make me look at the game from a different viewpoint. Kobolds of POWDER are like no other kobolds. They have their own society, own patron (Wpark the Wonderful) and own secrets. ADOM's rats also feel different to many other kinds of rat. Just read the description where you are surrounded by dozens of pairs of red glowing eyes and your torch dies ... On the other hand quote-based descriptions of NetHack could just as well not exist. Not helpful at all; not immersive at all.