random and balanced

[nurse]

Hi!

I've started developing my first roguelike. And allthough I understand there are no given rules of how to do certain things, from one thing I wonder what is common.

The question is basically:
Is a rl-dungeon is generated in a way there's always a way to win, but your decisions during the game can(will) make you lose instead.
Or are most dungeons unwinnable from beforehand, you just don't know.

Let's say in my game I want a food-counter and every step player loses one food-unit. I've generated a dungeon and now I'm gonna throw down pieces of food in it randomly.
Do I care if, due to the random nature, the first food to find is too far away to get to, even if you walk with the shortest path to it. Or Am I gonna place the food randomly but I check if the food is reachable if player takes right decisions?

What I'm intended to do is just throw the food in the dungeon complete randomly, and if, during testing, I die because of starvation too often, I will place a little more food. This will create unwinnable situations for sure every now and then.

I hope my question is a little bit clear and not too silly newbiesh. It's ofcourse not only about the food, but also about health-upgrades etc.

Thanks in advance!
 
 

[corremn]

You can do it any way you like.  I guess most people just randomly generate food where ever.  Thats what I do anyway.  A lot of roguelikes get over this by the possible chance go edible corpses being generated.

Starving at the start of the game could happen regularly and no one would really be that annoyed as long as it does not happen too much, (unavoidable) starving at end game would stop people from playing.

Generally you should use food as a clock to keep the player delving deeper, if they stay too long somewhere relatively safe they will eventually starve, if you force them to keep moving forward with the promise of new food around the corner then it is a much more enjoyable game IMO.  You could always generate some food on each level so if you were starving you could take the risk and go deeper.  If food and starvation is not meant to be used as a clock and serves no other purpose then remove it altogether.  Think about why you want food to begin with. Clock, corpse effects, magic mushrooms, fuel for actions etc.

You post has a deeper question about winnabilliy of the game.  Roguelikes have to be winnable full-stop. Well at least after a few minutes of playing.  If a game of solitare took 30 minutes to play only to then realise you could not win no one would play it.

[nurse]

Thanks for replying.

My post was indeed about the deeper question of winnabillity, food as an example.
Probably I'm thinking way to hard about this, but if you say a roguelike has to be winnable, do you mean: there must be a chance to win, or do you mean good strategic thinking has to make you a winner? In my point of view at the moment it's a game of chance with information, like poker.

If a computerprogram could play a game of rogue, and it runs through every possible scenario (this is theory), will it always find a scenario that's winning?, or is there a little chance that billions of possible scenarios will all end in death?
 
thanks again

[Quasist]

I think evry roguelike needs digital version of Dungeon Master. The AI that is not just moving NPCs. For example, detect if player does risky turns - like fighting high-level mobs or descenting deeper very early then help him/her with loot bonus or even special encouters

[Gamer_2k4]

Probably I'm thinking way to hard about this, but if you say a roguelike has to be winnable, do you mean: there must be a chance to win, or do you mean good strategic thinking has to make you a winner? In my point of view at the moment it's a game of chance with information, like poker.

I'm too impulsive to be any good at roguelikes myself, but I have the distinct impression that roguelikes can be won consistently by experienced players.  Winning a roguelike relies on knowing the game mechanics.  Once you know what's going on (such as needing a certain immunity before going below a certain dungeon level), you won't have to deal with unexpected instant deaths or other hindrances.

I would think that an easy roguelike allows the player to win once he knows how the game works, while a hard roguelike would require the player to make certain decisions even after the rules are known.  An example might be a dungeon where you have to deal with cold hounds and basilisks.  Do you wear the ring of resist cold or the ring of resist petrification? Basically, even when the player knows what resources he needs to win, the game will limit those resources.

[Anvilfolk]

Relating to food, I actually really like Crawl's version, even though it makes me grumble every so often.

Basically, you can only eat raw meat if you're Hungry or more. There's also Very Hungry, Near Starving, and Starving. Raw meat can be obtained by butchering corpses, although some isn't edible. It's a good idea to keep as many of the rations as long as possible, so this creates some tough tactical choices. If I'm getting hungry and level is still relatively unexplored, I find myself going as far as "Near Starving" before consuming a ration, exploring all the while. If the level is mostly explored, then you have to weigh whether you're willing to go down to the next level hungry and to tougher monsters. The orc mines, for example, are an abundant source of food, but really quite dangerous.

Generalizing though, I believe that the almost purely random approach is OK. Game mechanics and tactical knowledge should give you a much larger advantage (strength is better for maces, dexterity for spears and swords, kobolds can't be eaten or you'll get sick, certain levels are known for certain monsters which you can attempt countering such and such way), but should not guarantee success. Crawl has a way of constantly giving you challenges, whether they be strategic by making big, long-lasted decisions (which skills to focus on, dungeon branches with themes) or tactical by shoving monsters or groups of monsters your way that are definitely your match.

So basically, random is OK, but it should always have implicit input from the player. You should not say that level 3 always has this and that type of monsters, because the player can be really weak or really tough on reaching level 3. You can always chose to create more loot of some type based on the player's inventory, and this kind of stuff. If the player is too low on permanent food, you can generate more corpses but keep ration generation at the same level. This should, in principle, but not always, allow the player to restock on rations, IF HE PLAYS ADEQUATELY.

Unpredictability is a pillar of roguelikes, and so are random deaths you can't avoid. Knowing the game you're playing should highly increase the odds of survival, but never to the point or reaching 100%, or even 50%! I've never finished a roguelike, seldom go too far, no matter how hard I try or how many times I play, but I still have loads of fun!