Messages

1

Programming / Re: Realism in Roguelikes

« on: November 20, 2013, 06:54:05 PM »

While I realize that you're saying you're done with attributes, one other thing I wanted to throw in:

The current discussion's been focused on getting carried away with excessive attributes to represent different things, either for flavor or gameplay purposes, but if we're trying to split up all the different gameplay mechanics into different attributes to avoid bundling unrelated gameplay concepts together (e.g. it's annoying when your wizard can't carry anything because he has low strength because it doesn't help him survive otherwise), do we need "attributes" in the "strength/dexterity/intelligence/etc" sense at all?  Why not just have stats like "melee damage" or "carry amount"?

Part of this idea is coming from some games in a different genre entirely: League of Legends vs. Dota.  In DotA, you have three attributes - Intelligence, Agility, and Strength.  Strength boosts health and health regeneration, agility boosts attack speed and armor, intelligence boosts mana and mana regeneration, and one of them will also boost your attack damage depending on your character.  You can also get items that directly boost certain stats like damage or attack speed, but many items work in terms of attributes.

In League of Legends, the stats are just named directly after what they do.  There's no strength or dexterity or intelligence.  There's just "attack damage" and "ability power" and "mana" and "health" and so on.

So the question is, from a gameplay standpoint, do we need to name attributes at all?  I think one of the reasons having too many attributes can be a problem is that it can be hard to keep track of what each one does.  If you've got 7 different attributes, you have to know what they're all for in order to decide which ones to raise.  But it's much easier if one of your attributes is just called "attack damage".  There's no question of "wait, what exactly does wisdom do again?  Will that help on this character?"  Just "I need more damage, I'm gonna put this point into that."

You lose flavor in the process, of course, but you get transparency.  Some would accuse LoL's stat system of being overly simple or lacking in any flavor, but one of the things I really like about it is that it's extremely transparent.  The effects of each stat follow very simple, easy-to-understand formulas.  Even armor and magic resist function in such a way that makes it really easy to estimate off the top of your head how much difference they make (physical damage taken is damage/(100 + armor), and same for magic resist and magic damage).

2

Programming / Re: Roguelike Gameflow - Alternatives

« on: November 15, 2013, 09:53:26 AM »

Based on how flexible your system is, you could also try to work with a world that's consistent between characters, but where the story evolves over a certain number of games, so you do get to see how the plot finally unfolds, even if your heroes are invariably killed by rats. The success of a certain character would be measured in how much impact he/she has on the game world. Between each character, the procedural world/plot would advance a certain number of steps. So if one character is taking part in plots surrounding the crown prince, your next character might enter a world where the prince has already been coronated, or conversely one where some other faction has managed to assassinate or exile him.

I like this idea.  Persistent world, new characters.  You'd have to figure out how to get the right balance between the freshness of creating a new character and starting from scratch in most roguelikes so that things still feel "new" with new characters, but it's still a cool idea.  In a way, player Ghosts in DCSS are like  a tiny version of this idea - they give the sense that you're exploring the same dungeon all your failed characters died in - but it would be interesting to make something that worked like this on a much larger scale.

Isn't this a little bit of the idea of Rogue Legacy?  I haven't played it, but I know the concept is that every character is supposed to be the descendant of the last character.

Here's something else related to what we are talking about.  It’s the opposite of my previous idea in a lot of ways, but I like it too.  In ADOM, there is one situation where the PC's decisions actually determine what reality is generated, and it's quite fascinating.
[...]
So, the point is, the player could participate in actually writing the story, in deciding who was good and who was evil.

That's a neat idea.  Lots of RPGs let you influence the plot with the decisions you make, but it's always your actions changing the course of events.  I don't think I've ever seen games where your actions affect reality itself.  I like the idea of the game figuring out what you're expectations are and then deliberately conforming to or denying them.

3

Programming / Re: Realism in Roguelikes

« on: November 13, 2013, 08:47:41 AM »

I think I disagree with the notion of increasing the number of attributes being the best way to increase realism.  As you pointed out, it makes some sense and could be cool if done right, but would be hard to balance and in the end would likely add a lot of complexity and bookkeeping without actually adding a ton of depth to the game.  Sure, you could have separate strength and dexterity stats for each limb, or seperate agility and dexterity stats, but creating compelling gameplay from either of those ideas would be difficult.  You could also do things like have a food system where you actually have to worry about your character's nutrition and not just hunger, but a lot of people find hunger-only systems to be a nuisance, let alone a nutrition system.

I think to some extent, as Krice pointed out, you have to accept that a Roguelike will always be a simplification of things.  You're essentially reducing your character to set of numbers in any turn-based roguelike, after all.

However, it might actually be an interesting idea to try to represent your character's attributes as things other than numbers.  Consider health, for example.  It's not at all realistic that your health is represented as a number, and that the number being low is, in most games, meaningless, except that when it hits 0 you die.  If you really wanted to make a game more realistic, you could try to have a more realistic injury system.  Of course, the downside is that a fully realistic injury system would involve your character having to recover in bed for weeks after suffering a bad sword wound and possibly never being able to fight again, so you still have to find a balance.

One thing I think could be interesting would be trying to implement a "realistic" magic system.  I realize that the phrase "realistic magic" is silly, but what I mean is trying to create a consistent mythology of how the magic in the game works and actually implement the gameplay that way.  A lot of books have magic systems akin to a set of scientific laws governing how magic works, what it's capable of, etc, but most games just represent spells as things that drain from a meter that recharges over time.  It would be cool to see a game that actually has a complete mythos behind the magic system with a more complex end result than "spells cost mana and you can't cast spells when your mana's empty."

4

Programming / Re: Roguelike Gameflow - Alternatives

« on: November 12, 2013, 10:45:46 AM »

When it comes to a randomized story system, what about just randomizing the certain "intentions" of certain prominent NPC's? NPC's would always have the same alignment, like chaotic or lawful, but whether they were good and honest or evil and dishonest would be the random part (remember, I want to separate how chaotic a character is from how moral they are).  The probability might be weighted so that a character was usually “who they were supposed to be” but without it being a certainty. 

Since you could never assume an NPC was honest, speech skills that could detect dishonesty would be important.  Also, if the PC was desperate, they might rationally decide to take a chance and trust a normally dishonest character.  This would make sense, because it is possible that character had been marked and honest that play-through.

The idea of lying to the player in various ways is an interesting one to me in general, actually.

I've always loved the idea of games that lie to the player, and I think your idea of a Roguelike where the characters are somewhat fixed but their intentions vary is awesome.

If you designed the game around it, you could almost end up with a sort of whodunnit-esque scenario.  It would be like a Roguelike mixed with Clue.  You never know who the villain's gonna be each time you play it.

I think the main problem with this is how you stop permadeath from ruining the fun (assuming you've got permadeath).  You're not invested in the plot of your character, you're invested in the gameplay of your character, so when you die and your character's gameplay dies with it you're okay.  But imagine if you've spent hours unravelling the mystery of who the true villain behind the story is, and then, when you know you're getting close to uncovering it, you mess up in combat and get killed?  That would be frustrating, because a gameplay mistake made you lose your story progress too.

Of course, this is a much bigger deal if the whodunnit mystery is a deep plot you get involved in, and not just a giant puzzle in disguise as a story, and having an intriguing procedural plot is incredibly ambitious, but even if it's just a puzzle in disguise, losing your progress on solving a puzzle because of a mistake not related to the puzzle would be frustrating.  It would make sense to lose your progress if you messed up and killed someone who's not actually the villain, but it would suck to lose your progress on something unrelated to combat if you die in combat.

Another possibility is you could make it a normal Roguelike most of the time, and have plot twists be a rare sort of Easter-egg-ish element.  Most of the time, you've got a generic preface where the king tells you to kill the evil wizard, and you delve into the dungeon to kill him.  Maybe you get info throughout the dungeon about his evilness - scraps of paper describing his evil deeds, or he shows up for some mustache twirling, or whatever.  But maybe one time, instead, you find scraps of paper hinting that the wizard's just a scapegoat for all the horrible things the king has done with his tyrant-like rule.  The Wizard shows up and urges you to join him, but it's not the usual villain mustache twirling trying to turn you to his evil agenda.  He really means it when he tells you the king's horrible and he wants your help.  And you can accept, and then you turn around, leave the dungeon, and go kill the king instead.

This could still run into issues that would have to be addressed.  One issue is, how rare do you make the event?  If it's too common, then it loses it's specialness.  If it's too rare, not enough people get to experience it.  It also has issues with permadeath.  What if, after dozens of regular playthroughs, you finally get the plot twist playthrough, and then you die?  You just missed your rare chance to experience the special version because you happened to have a bad game that time.  That would suck.

So I think one of the questions that arises is, how do you combine procedural plot generation with permadeath?  It's one thing to lose your character from a gameplay standpoint, it's another thing to never get a narrative resolved.  You can try softening the permadeath so that you can lose a character without losing the narrative, but then you also lose a certain amount of the point of a procedural plot in the first place - that you can keep creating different characters and get a different plot every time.