Permadeath and Balancing Frustration vs Challenge

Hi all,

I'm working on a project called Dungeon Code (Devlog: https://forum.unity.com/threads/dungeon-code-castlevania-meets-nethack-video-based-devlog.926225/ ) which revolves around auto generated dungeons that are explored and conquered by the player.

My feeling is that death needs to be permanent, despite being a platformer the games main inspiration is NetHack, but I also want to avoid the constant save and retry that is found in NetHack.

(Most recent video for a little visual context)
​

Let me dump some of my thoughts:

Tough Luck Death is Permanent - This will add weight to conquering a dungeon making it more of an achievement.

If it works: Players get huge amount of play time out of each dungeon as they carefully work out optimum paths and get better at execution.

If it doesn't: Players get frustrated quickly and stop playing.

-=-

Save when you Descend - The dungeon is broken in to layers/levels, so a natural save point would be saving when you descend down a level.

If it works: Players get a good amount of replay time out of each level, but also know that once they beat some hard section and get down to the next level that they are 'safe'.

If it doesn't: It is still too hard and players get frustrated and stop playing or it becomes too easy and conquering a dungeon loses its weight.

-=-

Save any time - As with save scumming in NetHack, save at any time, save before a difficult section, replay as needed, etc.

If it works: Players feel safe and spend more time in the game. No permadeath frustration.

If it doesn't: All challenge removed from the game, players save before every obstacle. Becomes very bland.

-=-​

For some context:

• I imagine each dungeon would take about 10 minutes to solve if you knew exactly what to do. Most games will take longer as players have to explore and work out puzzles.
• The platforming element will not be that hard. Its definitely not like super meat boy where you would die over and over.
• All traps will have some kind of cue, either visuals, audio, or both.
• In summary: it should be possible for a careful player to get a level on the first try.
• I see this as a PC game (at least get it right on PC before looking at other platforms).
• Clarification: The term level here is used to mean 'level of dungeon', i.e. like a floor, rather than the typical meaning of stage/level.You can go back and forth between levels as needed.

----

Of course I will try options and run through play testing, but I would really like to hear thoughts of others, both on my options and also other ideas for balancing challenge and frustration in reards to permadeath.

Thanks for your time!

 

If there are puzzles to work out, I think you pretty much have to save at the start of each level. That's standard behavior, and for good reason; having beaten the first 12 levels, if you make a mistake on the 13th and have to start over from square one, most players are going to say "nope, I have better things to do."

Incidentally (and unhelpfully):

I'm pretty sure you meant "cue" here, but I love the mental image of traps with a queue of adventurers just waiting their turn.

 

'cue' ... hah good catch

The key idea here is that there are auto generated dungeons uniquely identified by a code (hence Dungeon Code). A single dungeon might be 5-6 levels deep and have 2-3 puzzles, and 1 or 2 quests.

Clarification: The term level here is used in the nethack sense, i.e. like a 'floor', rather than a stage/level. You can go back and forth between levels as needed.

You would have to redo puzzles, but there would only be a small number across all levels of the dungeon.

Still agree this could be tedious.

---

In the game, the distinction bewteen puzzles and quests is:

Puzzles - limited to one level, involve operating levers, switches, and using items (typically to open something).

Quests - span multiple levels of the dungeon, involve dialog, generally involve finding something or killing something.

 

Another idea I was thinking about was resurrection scrolls (i.e. Lives). This gives you an out from a one time mistake while still keeping the tension high.

 

I guess you need to ask yourself what the player gains from restarting right from the start.
if you simply want to make it more challenging, that might be frustrating to most people.
If they have a chance to redo it a different way, meaning they have better items and more health when they reach their previous death point, then they might be happy to try again.

imagine fighting a boss with 3 phases. You lose 1/3 hp on phase 1, 1/3 hp on phase 2 and then you die on phase 3.
Restarting from phase 3 with 1/3 hp would be kinda annoying - you'd probably want to go right back to phase 1.
However, if your health regenerated between phases, repeating phase 1 and 2 would be tedious and pointless.

 

The player can gain at the meta level, for example they can gain "Mastery" (through repitition), and they can gain a larger sense of satisifcation when they do finally complete. They will also have the opportunity to discover new things, but only if they don't play the same way each time.

---

The thing is the game will reward careful attentive play: a reasonable player who listens and looks for the cues should die very rarely. The flow and feel of the game will be orchestrated around this concept. I think there needs to be some significance to death to enforce this.

 

If you feasibly can, I'd leave it up to the players to pick their desired save settings in the difficulty settings. Many feel strongly about these things and they want to get different things out of games. Some simply won't play anything with permadeath, some won't play anything without it. As player skill increases, the preferance for a saving method can change too. Save anytime is great to learn and get better, permadeath is an added thrill once they are good enough to not be frustrated by it. Also this eases the balancing burden on you a little. If it's mandatory permadeath, you have to spend more time on balancing to make sure you get it right for a big enough chunk of players. I still think for example teleglitch is objectively too hard for a permadeath game.

 

I would take the Paradox approach. Create two different tiers of gameplay. The "normal mode" is more forgiving, allowing saves almost anywhere. Then a separate "ironman mode" which is far more strict, and is required for unlocking achievements. For example, in any Paradox game when ironman is enabled the game autosaves regularly and overwrites the previous autosave, so you can't go back to undo a mistake. (they probably only implement saves at all because a typical Paradox game play through lasts around 30 hours or so)

You could also look at Doom 2016's Ultra Nightmare mode, where in the normal skill levels you can save and go back whenever, but in Ultra Nightmare the enemies are turned up to the extreme and you don't get to save at all. One life, one play through, good luck. It is similar in concept to the Paradox approach.

 

I don't know how this can work with the competitive element of the game. If you conquer a dungeon first, you get to name it. I think this will be a signifcant driver.

And to be blunt, I'm not sure I like the idea of player agency here. Specfically I'm worried that if the player is nurtured too much the game loses the idea of 'mastery'. Of paying attention to the level, of learning the mechanics, etc, etc. So people might be happy they didn't die, but may also quickly get bored.

A hardcore/ironman/nightmare mode could work, I think what I can do is embed the difficulty in to the dungeon, for example instead of a 'hardcode mode' I can have a 'hard core dungeon'' (maybe 1 in 10 codes leads to a hard core dungeon for exmaple). If you hit upon one of these you could get a pop-up:

"This dungeon is HARDCORE progress cannot be saved, continue?"

[WIMP OUT] [BRING IT ON]
​

This keeps things fair, makes it clear what people are getting in to, but also allows for this kind of hardcore play. Defintiely worth some thought.

---

PS Thanks for all the input guys. I kind of feel bad disagreeing with people when they are taking time to contribute, I hope it doesn't come across as negative. I do have a pretty strong idea of how I want the game to feel, but also fully appreciate I may be completely wrong!

 

One danger of randomizing the "hardcore" dungeons is that people who're into that might not want to play the rest (and visa versa). You're forcing both players to engage with something outside of their interest.

 

It could be annoying having to keep hitting regenerate to find a hardcore dungeon. I guess simply simply triggering the part of the code that determines if the dungeon is hardcore over and over (behind the scenes) would do the job of allowing a player to filter. Not sold, but definitely worthy of consideration.

 

Something else...is the only thing that will differentiate the two types whether or not you can save? Or are you planning to change the design in some way? A name like "hardcore" implies there's more at play than simply whether or not you can save. And I can see that being a point of criticism - "These levels are supposed to be hardcore but they're the same as the rest."

 

I would look at things like: increased enemy damage, reducing number of key items (e..g no weapons in the level), enough to make it a clear challenge. But its not been fleshed out at all yet.

 

Couple thoughts:

I am a "build mastery" gamer, but I only want to build mastery after I am thoroughly taken by the hook. This could take 10 hours or more. In other words, difficulty, mastery, procedural content - none of these are hooks.They are things that add life to a game which has hooked me, but annoy the hell out of me if there is no hook (or at least not one I am interested in.)

I don't like to be punished for experimentation. Half the fun of games is trying fun things without fear of consequence.

Punishment for failure is cool if I only lose temporary opportunity. Losing significant time is instant refund from me. Five minutes is significant time in game world, at least to me. In other words, if I fail and that means I wont level up as fast, or I dont get to fight a special monster right now (but can have the opportunity later), that is cool. If I fail and that means restart, no way. That's outdated design IMO.

Procedural gen is a word makes me suspicious. I'd consider hand building levels in order to find best gameplay balance, then retroactively design procedural system to recreate what you've identified as fun.

 

The game is intended to feel like NetHack, and I don't mind at all if that turns many players away, as long as it doesn't turn everyone away. To feel right there has to be consquence. For NetHack the only way to avoid permadeath was to go outside of the game to backup your save files, it was obviously cheating. Death there could you lose you days if not weeks.

I don't want to be that hard core (there wont be days or even hours of gameplay in a single dungeon). But I do want death to be something players care about.

There are plenty of hard unforgiving games in recent days, including some platformers, Hollow Knight, Cuphead, Salt and Sanctuary, spring to mind. I don't think its outdated to design games this way, its just become more of a niche.

The pieces are hand built (and there will likely be certain codes that lead to fully hand built levels 'you discovered a special dungeon'), but procedural generation is the heart of the game. I've certainly no problem with handcrafting levels during design to find and explore interesting concepts, but I don't want to lose site of the goal.

It looks like a platformer, and that might be what keeps people playing for the first 5 minutes, but its mostly a game about exploring.

 

Is this intentional? Are you planning to market it as a platformer when that's not the essential experience? Because that sounds like a great way to make people upset.

 

Mechanically it is of course a platformer but it is not Super Mario Brothers. It's not like there isn't precedent for games that use platforming as a mechnic to provide different kinds of experience, Faeland, Spelunky or Himno for example.

Edit: Yes you will need some platforming skill to play the game, yes there will be platforms, traps, enemies, etc (as seen in the devlog). But the game is not about high precision platforming or twitch-based reactions. Its about finding items, discovering secret passages, learning enemy patterns, solving puzzles, being attentive, building up your character to equip them for deeper levels of the dungeon, etc.

---
I think it is way too early to talk about marketing. Obviously the intent wouldn't be to mislead

 

Modern games with permadeath tend to have some overall level-up. When you die, your score is converted into "town expansion points" or something. Those can unlock new classes, or give better starting items, or simply add buffs. Even Darkest Dungeon is somewhat like that. They play well both ways. On the one hand, you need to get to level 20 and kill the mini-boss without dying a single time. On the other, you can make a quick run, dying when you die, grinding out more town points.

 

This is a really good point. I need to think about what can be done in this regards.

The challenge I have is around the idea of beating a dungeon giving the naming rights and how perks could make this seem unfrair. But if that advantage isn't of too significant a utility I think it could work.

Maybe unlockable classes which give both pros and cons.

Or maybe the grind is considered part of the challenge and thus wanting to name a dungeon adds more value to the grind.

Thanks, I really like where this is heading.

 

Letting players freely (even if gated by some challenge) name anything that is public, other than their account name, is a terrible idea imho. If you let them enter names for dungeons they beat, you'll have a wealth of literally the most offensive names people are able to think up. I'm thinking racist, sexist, antisemitic, homophobic etc.. And even if you try to be clever by only letting them use building blocks that you think are harmless, they'll likely find a way. Like in Dark Souls the countless "Try tongue but whole" next to corpses. Imho it's a huge mistake to let players freely name something that isn't just local to their own gaming experience.

I think a better option would be to for example place a statue at the entrance of a dungeon that has a plaque with the name of the player that first beat the dungeon. It'll still occasionally have the bad things I mentioned in there, because there are people with really offensive player names on steam, but the percentages will be way better imho.

 

I don't mind the statues/name idea at all, it is really a very good idea (also allows me to have more than one person get recognition, for example top three with different color/sized statues)... but why do you think the account name is any safer?

I can put a lot more constraints on a dungeon name (must be a combination of words picked from a limited list) than I can on account/display name.

 

It isn't safer at all. I was talking about the comparison to freely naming things. You didn't mention it being a whitelisted naming system before I think. If you are super strict about the whitelist, it's a lot safer indeed, but people might still try to find some meta way to mess with it.

Have you thought about using a thumbs up/down voting system for letting players tell you how much they liked a dungeon? That seems like a useful metric to collect and tweak your dungeon generation algorithms.

 

Was thinking of a rating of some kind (e.g. 1-5), but also detailed analytics which track stats like: dungeon starts, dungeon levels reached, dungeon time spent, dungeon completion, dungeon percentage explored, etc.

I'd also like to build some pipelines to see what kind of deaths cause players to stop playing a dungeon, what kinds of features cause people to play a dungeon longer, etc... but I don't know how deep I want to go. This is supposed to be a one man project

Cheers,
John A