Tuesday, April 24, 2012

Episode 32: Combat Mechanics

Welcome to Roguelike Radio episode 32. This week's topic is combat mechanics in roguelikes. Talking this episode are Darren Grey, Ido Yehieli and Andrew Doull. The mp3 of the podcast can be downloaded here, played in the embedded player below, or you can follow us on iTunes.

Topics covered this week include:
- Original mechanics in Rogue and their D&D inspiration
- Realism in combat mechanics (and how it doesn't always work)
- The game of numbers, and how devs love bell curves
- Combat logs and other ways of relaying combat data
- Weapons flavours and damage types
- Movement and placement around the field of battle
- Area effects of weapons and special attacks
- Bump to attack and the unification of movement keys with other game functions
- Deterministic vs random systems
- Status and timed effects

Some games highlighted:
- Brogue
- ToME4
- DDRogue
- Zaga33
- Sword in Hand

Join us next time for more exciting and information roguelike discussion!


  1. Another episode already? I like this change of pace...

  2. And it was recorded last night too.

  3. In IVAN you cannot attach an arm of another monster, but there is indeed an important feature of exchanging your limbs into ones made of powerful materials.

    I would like to advertise the unique combat mechanics in Hydra Slayer. There are weapons which are more or less powerful, but as a general rule, the more powerful the weapon is, the harder it is to find an enemy that you can use it on; also, for most weapons, some hydras will be resistant (or even made stronger), based on color. The system is deterministic, and based (mostly) on simple mathematical rules, so you have to do some planning in order to find the sequence of attacks which slays each hydra most effectively; sometimes it is even worth it to include attacks that seem to make the enemy stronger. To be powerful, you need to pick up (or create) a set of weapons of different types, power, and colors which effectively kills as many hydras as possible (you are allowed 6 weapons in the end for the coffebreak game, and up to 10 for the full game). Health regenerates only if you fight effectively and it is thus an important resource, you will eventually lose all your health before winning the game if your set of weapons is not good enough.

    Much better than a typical system, where you have lots of numbers (stats, accuracy, dodge, armor, damage, critical hit chance) which affect your effectiveness in unclear ways (all these bell curves are hard to analyze and you often even do not have the formulae), and you typically end up in guessing what is good for you, and using one weapon for everything.

    Overall, after playing games with well designed combat systems, the typical "bump to attack with lots of numbers" feels boring and needlessly complicated.

  4. You should do a show on Sil. I only discovered it recently. Technically it is an Angband variant, but a very, very drastic one -- it's more accurate to call it a reboot of Rogue that uses parts of the Angband engine. No town, 20-level dungeon, no character classes, no levelling, almost no magic. Excellent and innovative combat system that makes melee combat interesting eg there are skills for charging, attacking while retreating, attacking multiple opponents, attacks of opportunity, flanking attacks, knockback. Also excellent and highly viable stealth system. Experience is awarded not just for killing monsters but also for encountering them, exploring the dungeon, and identifying items. Several people have won the game with "stealth" characters that have never attacked an enemy.


  5. I think over simplicity in a game mechanic is preferable to over complexity. The more entropy involved in a performing a given action, the harder it is to reason about it and develop strategies. If there are ten different ways to attack all the time, one of them will probably be unbalanced, leaving the rest useless.
    I think there's definatly room for a higher level of simulation in combat (vats in fallout, IVAN's body part system ) but I don't think that bump to attack is inherently a bad mechanic just because it isn't complicated.
    I think exposing a large body of vaguely interconnected numeric values to the player is worthless and I wish less games did it. Computers were invented to allow people not to have to deal with big piles of numbers.

  6. 1) You discussed Dwarf Fortress as the only game which has made a reasonable attempt to determine combat statistics from physics/material simulation. DF has a very different goal from most roguelikes, though. Most RLs are using a fixed or mostly-fixed set of weapons and abilities to create procedurally-generated dungeons and combats. DF, at least in the long-term, is trying to allow procedurally generated materials, weapons, and fighting styles.

    In most games, it's massive overkill to look at a "well-made bronze longsword" and to start doing physics to figure out how well-smithed bronze distributes shear forces. At most, you might want that "longswords" are medium-sized cutting weapons that use the "sword" skill, that "bronze" gives a weapon -1 to hit and damage, and that "well-made" gives +1 to hit and +1 to resist breakage.

    Without simulation, it becomes the designer's responsibility to make sure that bronze works "similarly" in bronze longswords and bronze helmets. In a lot of games, bronze might just mean "better than copper but worse than iron". In more specialized games, maybe brittleness will become involved. But there really aren't very many metals in the world; it's a lot easier for the game designer to take on this responsibility than to try to write a research-level physics simulator.

    DF, on the other hand, is actually hoping to procedurally generate new types of metal, new types of armor or weapons, etc. And if this particular game world is rich in wonkatite ore, it needs to make sure that the properties of wonkatite swords, wonkatite maces, wonkatite plate mail, wonkatite ballista arrows, and wonkatite mechanisms are reasonably related.

    So I guess my takeaway point is that DF isn't just the only game that's succeeded at anything like a physics-simulation combat engine. It's also one of the only games that has any legitimate *need* for one.

    2) In your discussions of healing, I would have liked to hear references to Cataclysm's pain system.

    The Cataclysm character has hp numbers for each of six body parts (head/torso/arm/arm/leg/leg). Running out of head or torso hit points will kill you. These hit points are unrealistic in all of the traditional ways: they can be quickly restored by bandages and first aid kits, and there's very little in the way of permanent damage or combat penalties. (You'll temporarily lose use of a limb when it's very low on hp, and permanently if it hits zero, but these rarely happen except just before death).

    The real danger of combat is your character's "pain" level. Every wound increases your pain level, and these don't go away when you slap bandages on. Being in pain causes stat and speed penalties that can quickly become substantial. Deaths in Cataclysm usually happen when you start doing poorly in combat, and end up in too much pain to outrun or outfight your enemies. The only thing that decreases your pain level is time (and technically one of the very rare bionics abilities).

    The main way to get around the pain system is via painkillers. Taking a painkilling drug will cause it to slowly build up in your system, and then slowly decrease over time. Each point of painkiller decreases your effective pain by 1, but pain typically lasts longer than one dose of painkillers, and most of the painkillers have various levels of addictiveness and other unpleasantness to deal with.

    Anyway, it's a really interesting solution to the unrealism of the "no combat penalties for wounding, bandages are magical" systems. You can still do cinematic/unrealistic things in a fight, as long as you pop 2 oxycodone first; after the oxycodone wear off, you're either going to need to spend some time recovering, or you're going to have to keep popping more painkillers and deal with addiction. And it works really well in the "alternating exploration and fighting" gameplay that Cataclysm is aiming for.

  7. One roguelike where the spells can hurt you is Castle Of The Winds, where the "ball" spells like Fireball will damage you in the blast if you use them on a creature that's right next to you (though sometimes you'll do it anyways, if you have a resist or there's several creatures or whatever).

    Also, yes to IVAN. Lots of good stuff in IVAN. The game needs its own episode some time, even though it's no longer being developed.

  8. I think Dwarf Fortress' adventure mode is plenty fun, but it has a much different type of focus and scope to most roguelikes. Because unlike most roguelikes you aren't really trying to get to the bottom of a dungeon or complete a particular quest or goal. The old DF maxim of "Losing is Fun" is as much alive in adventure mode as in fortress mode. The players in adventure mode are ultimately disposable. Exploring and unlocking lore about the world are the real draw in adventure mode, not necessarily building your character and completing a goal, and the mechanics work well for that.

    That said, you do tend to underestimate how tough an adventure mode player can be. DF has it's share of players who made near unkillible adventurers who could punch dragons to death.