What do you hate most about Roguelikes?

[Jastiv]

Rogue-likes seem to be an incredibly popular genre in the free software world, with flagship games like nethack, almost everyone has played one, or at least tried to play one.
Also, a lot of people don't like rogue-likes for various reasons. What do you hate most about them?
Personally, I hate the tremendous difficulty of getting through the game with only one life. That is probably the thing I hate most about them. The second thing I hate about them is just the general difficulty of the games. If I had to add a third thing, it would probably be how turn based games lend themselves so poorly to multi-player, but that is another topic.
Wograld has some aspects of rogue-likes, but obviously not the stuff I hate the most.

[fluffrabbit]

I hate high-resolution text-mode graphics where the efficient nature of text rendering is exploited to put way too much shit on the screen for me to be able to focus on what's going on. It's not immersive, it's overwhelming.

I hate diagonal movement keys. The Legend of Zelda: A Link to the Past figured out how to do diagonal movement properly with only 4 arrow buttons, and that is the model I base my movement systems on. 8 directional keys feels antiquated, like something from a Colecovision/Intellivision game.

I hate that twitch reflexes are not rewarded. I know that's the central nature of a roguelike, but I hate it.

[dulsi]

Nothing. I don't have any problem with rules of a roguelike. I'm not fond of permadeath and used backup files to prevent it back in the day. I don't play them that much any more due to the time needed.

[domtron]

Not trying to fight or anything, but the question struck me as coming from an odd direction and I dislike it for two reasons.

A) It's rather negative, if that makes sense? The question is basically saying "lets complain about Roguelikes!" When I first saw it I thought you were angling for a "What do you hate about current roguelikes and how can we fix it."

B) Building on A, part of your answer in op doesn't make too much sense to me. It comes across like you just don't like Roguelikes at all. While the definition isn't nearly universally agreed on, many define the Roguelike genre as:

• Permadeath, losing all progress made during the game.
• Heavy use of procedural generation.

There are a few other things that makes up a roguelike, but if you drop those things your really not making a roguelike any more. Even when permadeath is essentially removed it still exists in some way (You have the option to turn it off, you lose some but not all progress, etc). The difficulty of one life is some of what makes it fun.

IMHO it's like saying you don't like managing different attack strategies or an economy in a strategy game. Or you don't like the speed/twitch reaction of an action game. If you don't like the core ideas of a genre then do you really like the genre at all?


If I was going to criticize current roguelike games I'd say too many games try to label themselves as a roguelike when they really are not. But that's just my opinion.

[drummyfish]

Never even tried any roguelike, mostly for one reason:

permadeath

I actually like the idea of ASCII/terminal interface, I'd play such a game only if it were more like The Elder Scrolls (open world, freedom, work on your character) rather than Tetris (get the highest score).

I have a suspicion the popularity of roguelikes stems from the simplicity of implementation (no graphics, no physics, no sound, no saves, no multiplayer, no hand-made levels, ...) rather than outstanding gameplay. (That's all okay but I'd rather see one complex TES-style game than 20 new flavors of roguelikes.)

[fluffrabbit]

TES games technically have permadeath in that you don't respawn. They just also happen to have a quicksave option so can un-permadie.

[dulsi]

I have a suspicion the popularity of roguelikes stems from the simplicity of implementation (no graphics, no physics, no sound, no saves, no multiplayer, no hand-made levels, ...) rather than outstanding gameplay. (That's all okay but I'd rather see one complex TES-style game than 20 new flavors of roguelikes.)

I believe most roguelikes include graphics at this point. You can still play with text characters but I'm not sure what is most common. Simplicity is not guaranteed of roguelikes. NetHack has so many different commands they are just a single letter like in Angband. Some of them have crafting/enhancement systems. Cataclysm: Dark Days Ahead has a skill system which looks quite daunting to me. It also requires SDL2_mixer so it probably has sound. I'm not certain no hand-made levels is simpler than procedurally generated. I mean a quick procedurally generated dungeon is easy but making one that is interesting to explore may require some work. They may also have rooms that are hand-made and added to the dungeon. Dwarf Fortress (closed source game) has been praised for it's depths and complexity. Have you taken a look at Gearhead series? I still have yet to try it but I've heard good things about it. No multiplayer is due to the turn based nature but there are still some like mangband.

They may not appeal to you but I think they have a lot more in them than you give them credit for. (I know I don't like FPS. Every once in a while, I try one because of the reviews or neat concept. I keep finding I just don't like those style games.)

[domtron]

A quick defense of Permadeath. :P

I like Permadeath in certain contexts. A building/design heavy game like say Mine-craft/test is not a good place for permadeath. The game needs to actually be designed for permadeath.

Permadeath makes game progression more dependant on player skill. I think it increases the pavlov effect (I think I'm using the right term). You do X or do Y poorly and you have a harsher punishment, but do things right and you make it farther than in the last run. Each hit point is more precious because it brings you closer to a permanent end. It makes the player do things more efficiently each time so they can get farther.

Permadeath is also good to set a certain tone for a game if they are trying to make a point. Can be used for either philosophy of "nothing matters everything will reset anyway" or "make choices carefully or suffer a dire consequence". Probably others too, I'm no expert writer able to come up with those things.

[fluffrabbit]

I think permadeath presents a philosophical issue about game design.

Let's say you design the most realistic medieval knight simulator ever. The knights go into battle surrounded by troops in heavy armor. The actual likelihood of death is quite low, despite appearances. If permadeath is defined as an autosave system that deletes everything when you die, there is an actual risk associated with making creative decisions. So tone matters. It's a story.

On the other hand, roguelikes with permadeath are pixels and symbols that nobody cares about, so imho they're just arcade games rather than life sims.

[drummyfish]

I guess permadeath is a topic that can be discussed endlessly from all different perspectives, and I am not saying it is bad per se, just that it's what keeps me from playing roguelikes.

My point is I like RPGs (character development, complex mechanics) and I like Tetris (permadeath, no saves, simplicity, skills and highscores). I just can't for the love of god find the enjoyment in the mix of these two.

[fluffrabbit]

I like arcade games like Metal Slug and Street Fighter, and roguelikes feel like arcade games. Except you can play your fill of arcade games in an hour or two, while roguelikes can go on and on and on, and when you die it's over. If you only play games to kill time, I suppose roguelikes are pretty optimized for that task.

[drummyfish]

Or another take: it feels like a big waste to create a complex game (e.g. as you mentioned the alchemy system etc.) and then make it an arcade. An extreme case of this would be e.g. creating something like GTA V, then adding permadeath and only let people compete in how far in the missions they can get without dying -- you kill most of the potential fun with such a complex game -- the endgame fun -- and probably put off a large number of potential players. It's okay to add an arcade mode, but it shouldn't be all there is. I already feel the same about story-driven games that simply end as soon as you finish the story and don't let you play any further, it's a wasted fun.

Also please only take this as an attempt at a constructive criticism of the genre, it's great that many people enjoy it, but this is something I feel a lot of people might be missing. Someone please create an ASCII RPG in TES style.

[fluffrabbit]

I have talked about alchemy systems in other threads. There is much work to be done on that front, and my current proprietary game project completely revolves around the alchemy system. But for roguelikes, in my limited experience with the genre, it seems that developers shove in tons of loot to collect and other things. It's more than just alchemy.

If we're talking Nethack, it's an RPG that seems to be as complex as Dungeons and Dragons, but one can get blasted into oblivion so fast that I have derived no joy from it. It's a big bag of content that you can't enjoy; you just get killed in the second room if you don't know what you're doing.

Someone please create an ASCII RPG in TES style.

There are roguelikes that are purely graphical. The best example I can think of is the open source Android game Pixel Dungeon. It's written in Java and the desktop version isn't FLOSS, but it's still an okay example visually. To each their own.

Personally, I try to imagine myself in a future where you jack into a virtual reality and have your memories erased, and the more convincingly real the experience is, the less likely your brain is to forfeit it. As Agent Smith said, humans measure reality by the amount of suffering. I'd rather suffer realistically than screw around with unconvincing text graphics and respawns, but that's just me.

[Lyberta]

Someone please make roguelike first person shooter.

[fluffrabbit]

Someone please make roguelike first person shooter.

It's been done. One is called Superhot. Its roguelike element is time, though I believe the levels are manmade.

There was another one someone was making in Unity a few years ago, a procedurally generated dungeon crawler, though I think it was realtime.

[drummyfish]

Superhot was awesome but I can't see how that's roguelike. Anyway, we need a free clone please

[fluffrabbit]

Superhot was awesome but I can't see how that's roguelike. Anyway, we need a free clone please

It's a roguelike because it only goes when you move.

Anyways, here you are calling for the commencement of a 3D FPS project that would probably take a lot of time unless it were a mod to an existing open source FPS like Sauerbraten. At the same time folks around here don't want to wait for the development of a simpler and less ambitious 2D game clone like Tux Torture. (A truly serendipitous name imho.)

[Lyberta]

I just want FPS with random maps so I play it 24/7 and not get bored.

[fluffrabbit]

Procedural maps that aren't boring? That's non-trivial. Minecraft/test use Perlin noise and large open spaces, but I think they stay interesting just by virtue of coming with a lot of content. There are Mine-clones in the FPS genre, but they're the epitome of cheap garbage.

[drummyfish]

To me procedural environment is for some reason practically always boring, not really sure about why -- it's as if it's missing the life, the possibility of being surprised by something only a human can create, like a little reference, something funny, clever. Though the world can be unimaginably huge, you see through the pattern very quickly. I am often thinking about how this could be fixed. Maybe semi-procedural could be the answer, i.e. start with a huge procedural world, then start making changes to it by hand. Actually that is what players do in Minetest, except it would be done by the developers.

I've been once thinking about an experiment, a completely procedurally generated game, including its genre, rules, all the assets etc. Like 4294967296 in 1. Each game would be completely defined just by the RNG seed. I think it could be fun to just explore these games. But someone would have to implement it first.

[fluffrabbit]

If the lack of human intelligence is what bothers you about procedurally generated content, that's an AI problem. It should be doable. If a small number of bots with human-level intelligence changed the procedural world much as humans change their natural environment, the experience would be authentic. If one or more bots had godlike creation powers, the experience would be more game-like.

On the other hand, procedurally generating the genre and rules and just letting the computer do "whatever" doesn't sound like fun to me. Even with human-designed games, you can get a pack of 10,000 NES roms and play all of them, and 99% will be shit. There has to be some direction to make the game not shit.

[domtron]

Procedural generation is often used as a lazy way to build the world with less effort, but good PG can actually be harder then doing it by hand. The problem with PG in most games and what makes it boring is the lack of detail.

Lets compare the beginner programmer dungeon crawler, Minecraft, and Dwarf Fortress.

The beginner dungeon crawling roguelike generates the dungeon. The common algorithm picks a number of rooms, varies their size, and dumps it into the map then connects everything with corridors, and finally places start/exit and some monsters. It is kinda boring because you have very low detail. You have walls, ground, rooms, hallways, different sized rooms, and a few entities running around. Add in loot and it get more interesting because you have more to find.

(Side note: not bashing rougelikes here just using a simple implementation of one as an example of low detail PG.)

Minecraft expands on the detail. PG encompasses biome placement and size, terrain height details, twisty caves, decorations like grass and varying colors across biomes, some structures(they are premade for the most part, but placement is generated), loot, and enemies. Mechanics complement it since instead of passing through you take what is given and morph it into more. But it's boring in that there isn't all that much variety. Only a few mob types and they are not smart. The buildings are mostly premade and just PG placed. The few buildings that are PG don't have a real plan like nether fortresses where they run into themselves, suddenly end, and feel random in a bad way and very, very samey.

Dwarf Fortress is kinda the king of detailed PG for now. It generates not just terrain but history, items, and addon details of creatures. Things and people have history and personality, the very world does as well. Terrain is generated from a top down approach where things are placed in a way that makes sense. Mountains rise in the middle of land and the rivers flow from them to the seas. Settlements are placed in reasonable locations and connected in a fairly believable way. Generation of actual cities is a little rough but not worse then other examples to my knowledge. It would be even more interesting if it was visually varied as well.

Anyway my point is PG benefits from more details working from general wide sweeping ideas down to the little details. At least that's my idea on what's often wrong with procedural generation.


I also want to point out that procedural generation is not equal to random generation. Procedural generation uses randomness to an extent, but requires significant design (human intelligence) work to make sure parts come together as seamlessly as possible. Good PG will throw out randomness that doesn't match up to the bigger plan. Random is completely incoherent with the occasional happy pattern, procedural follows a big picture plan with variety choices provided by RNG.

Good Procedural generation in my opinion is layered generation. You generate general aspects and zones for the world, then layout features on the continents, then generate specific sections of the continent, and finally add in the first person details like that flower or that animal den.

That's my opinion/experience anyway.

[fluffrabbit]

I know of standard algorithms for generating terrain and dungeons. Those things are a science because said structures have uniform patterns IRL. As for placing stuff in the world (loot, mobs, etc.), that seems to depend on skill.

[drummyfish]

There has to be some direction to make the game not shit.

Direction would be procedurally generated

The problem with PG in most games and what makes it boring is the lack of detail.

PG can do infinite detail very easily, I think it has to be something else. Maybe it could be that the "interestingness" generated by PG is extremely sparse, while hand-made art has it very concentrated, because a human discards uninteresting stuff and only keeps the good things in, so that the result is very attractive to other humans.

There is this program, Manelbulber, for exploring 3D fractals, which is basically PG. I like to play around with it, the fractals are literally infinitely detailed, complex on the large scale as well as on the smallest scale... but still I've had more fun exploring much less detailed, comparatively very tiny worlds in open world games, created by humans. Even such worlds as the early Pokémon games with people being represented by a bunch of pixels, or even books with only words, felt more exciting to explore than the infinity of fractals.

[fluffrabbit]

Direction would be procedurally generated

That sounds to me like an AI-complete issue.

interestingness

You mean novelty?