Musings on an alternate timekeeping system

[Derakon]

Let me just preface by saying I'm not going to be implementing this anytime soon -- it's solidly in variant territory and Pyrel needs to hew to Angband's base mechanics pretty closely if it's to be accepted as the successor. But if it the misty future I happen to have the time and inclination, one of the things I'd like to do is overhaul how time and combat interact with each other.

Long story short, I'd lift wholesale the combat mechanics from the Grandia series of RPGs. In those games, there's an "action bar", and each actor in the fight is positioned somewhere along the bar. The bar has a "Wait" section and a "Charge" section. When an actor gets to the end of the Wait section, they get to choose what action they want to take. They then must pass through the Charge section, and when they get to the end of that section, they actually act. Then they're sent back to the middle of the Wait section to start their "turn" over again.

So far so simple. Functionally this is very similar to practically any other turn-based game. But you can do a lot with this framework:

* Different actions take different amounts of time in the Charge phase. Moving or attacking take no time to charge; using items is moderately quick; casting spells or using special combat techniques can take much longer. Charge time can be reduced by training the appropriate skill.

* You can see what different actors are planning on doing, once they get to the Charge phase. Whoops, looks like that ogre's planning to use his smash attack on me; I'd better defend. That dragon wants to breathe fire on me; I don't have time to cast Resist Fire but I can quaff this potion I have.

* Different actions also require different amounts of recovery time after taking place (by putting you closer or further back along the Wait bar). If you decide to defend, then your next option to take an action will come up quickly because you didn't actually do much of anything (just change your stance for a bit).

* Different creatures pass through the Wait bar at different rates. Bosses tend to get turns very quickly...but their more powerful techniques still require significant charge time, giving you the possibility of being able to do something about them.

* Perhaps most importantly, you can manipulate your opponents' positions on the action bar. Some attacks have a "Cancel" effect, which moves the target back along the bar -- and the degree of cancellation increases as the target is closer to actually taking their action. The effect during the Wait phase is pretty minimal, but an actor in the Charge phase will lose almost their entire turn (getting sent back to near the end of the Wait bar).

You get some neat things out of this. For example, casting an area-of-effect spell on a single target, because you think that by the time it goes off, it will hit more enemies. Or there's an enemy moving towards you to hit you as you get your turn -- do you have enough charge time available to use a special cancellation tech, or should you just defend? Your ally is about half a second away from getting hit and won't get their turn in time; use an attack item and hope it goes off in time or try to run over to physically hit the enemy? A dragon's getting set to breathe fire on you; do you have time to get out of the way, or even to chug this potion of Resist Fire?

Applying this to Angband is actually not too difficult. You'd need to introduce the charging time (currently all actions are instantaneous on getting your turn), and you'd need for different actions to cost different amounts of energy (which we already have but we don't message it well). Most importantly though, you'd need the action bar; the system is no good if you can't see what your enemies are planning and adjust accordingly. Representing this in ASCII is tricky but by no means impossible...so long as the number of actors is reasonable.

Of course, a system like this would not work well when there can be dozens or even over a hundred enemies involved in a given fight. It works much better when there's 1-5 player characters and a similar number of enemies. I think the most chaotic fights I've played in the Grandia series had at most 8 enemies (and 4 PCs). This system would probably also be well-served by shortening the range across which combat can occur in -- there's not much you can do about dragonbreath if the dragon in question is 20 squares away.

[Magnate]

Wow. That's a truly massive change - a far bigger shift than anything I've ever envisaged. It's right up there with persistent levels IMO. Angband is currently like a multi-actor chess game: you move, they (all) move; you move, they move. Adding the action bar makes it a totally, utterly different mental challenge. It could be a lot of fun, but my immediate reaction is that it's better suited to real-time play (in fact something very like this happens in WoW) and could be easily exploitable in turn-based play. But I've never tried Grandia and it obviously works there so who am I to judge.

[Derakon]

Well, Grandia is real-time with pause, much like, say, Final Fantasy. The main differences IMO between it and a classic JRPG combat system are:

1) Position on the battlefield matters. Some techniques hit an area instead of a specific target, and if you're far away from an opponent then they can't reach you to use standard attacks.
2) You can see what an opponent is planning on doing, before they actually do it.
3) You can cancel enemy turns.
4) You can improve charge times on abilities by training in them.

I think you could actually implement this in Angband, except that pausing would also happen whenever there's a -more- prompt (or while an animation plays out). The way I'd envision it working is that the action bar would be a vertical bar along the side of the screen (giving ~24 increments to work with). Each actor would be placed along the bar based on how much energy they've accumulated, and they'd have a colored dot indicating how close they are to moving to the next increment (i.e. red when they've only just reached the current increment, green when they're just about to move to the next one). Then instead of your next turn happening immediately after your previous one (assuming no messages / animations occur), you wait a bit and watch the other actors move up the bar and execute their turns.

The net effect is that you can get a pretty good feel for how quickly an actor is moving through the bar while still maintaining that ASCII feel. Something like this:

ACT - p             The Swordsman hits you. -more-
 | 
 | - D breathe fire
 |
 | - @ cast Resist Fire
WAIT
 |
 |
 |
 |
 |
 |
 |
 |
 | 
 |
 |
 |
 |
 | - o - y
---

The '-' next to each entry would be the bit that changes color based on progress. If you had, say, 8 colors then there'd be almost 200 increments to work with, which seems like plenty.

[fizzix]

I was thinking a bit about how such a system could work in Angband, and my conclusion is that it would look like a very different game. I like gedanken-experiments like these, so let's go follow my brain as it traverses this new idea.

All the games I can think of that use a mechanic like this are many vs many. Either the player controls multiple characters (like in Grandia, or the example I'm more familiar with, Final Fantasy Tactics), or there are multiple players (WoW). So the question is can the mechanic be interesting in a 1 vs 1, 1 vs few, or 1 vs many game like Angband. I'm going to ignore the tactical decisions that Angband already has, mainly, "should I use or save this item?" and focus instead on what new tactics the timing system adds.

I'm going to assume that both player and monster actions have some delay between the selection of the action and the execution. Furthermore, in the time between the action being selected and being executed, all monsters or characters can see what the action is, provided they can see the individual in question. I draw three minor conclusions from this.

1. There must be actions with a long enough execution time that the enemy has a reasonable chance to get a "what do you want to do" prompt before the action is executed.

2. There must be actions with a short enough execution time that a character/monster will be able to interrupt or react to the monster/character's action.

3. There must be more than a linear relationship between the power of an action and the execution/rebound time. If the relationship was linear, then it's always preferable to choose the shorter timed action.

I would argue that this mechanic doesn't create thrilling tactical situations in a 1 vs 1 situation. It may involve calculating your actions so you have the maximum time available to execute something while the opponent is going through its motions in order to interrupt it. I'm assuming of course that monsters are dumb and choose random actions regardless of tactical merits. If monster are smart, oddly, things get more boring. The best action will always be the most powerful action that can be performed before the enemy has a chance to react.

In 1 vs many things get more difficult. If you make the assumption that monsters can and will interrupt a player action if possible, then you will nearly always be limited to actions with very short execution times. On the flip-side the player won't be able to interrupt very monster, so monster are free to choose powerful actions with long execution times with the knowledge that the player can't stop all of them. So 1 vs many is out of the picture as far as interesting tactics are concerned.

In the middling cases, 1 vs 2 or 1 vs 3, you might be able to get some interesting tactics, so this is my first main conclusion. We'll call this the 1 vs few situation.

Conc 1. Timed-Tactical-Angband would have to be redesigned to be predominantly a 1 vs few game. Monsters would come in small packs of 2-4 and the player would usually engage all at the same time.

There's yet another wrinkle. If you add tactical complexity to the game, it means that the player is going to spend more time on each battle. In the case of Final Fantasy Tactics you're looking at 40-60 battles for the entire game, each one lasting 10 or so minutes. This may be at the extremes as far as tactical combat goes in your consideration, but even if we remove the time it takes to get through a monster encounter, you still need to cut down the total number of encounters to keep the game playing time at a reasonable level. So that's the second conclusion.

Conc 2. Timed-Tactical-Angband would require less total encounters in rough inverse proportion to how much more player time is spent per encounter.

Anyway, this was a fun topic to think about.

[Nick]

There's yet another wrinkle. If you add tactical complexity to the game, it means that the player is going to spend more time on each battle. In the case of Final Fantasy Tactics you're looking at 40-60 battles for the entire game, each one lasting 10 or so minutes. This may be at the extremes as far as tactical combat goes in your consideration, but even if we remove the time it takes to get through a monster encounter, you still need to cut down the total number of encounters to keep the game playing time at a reasonable level. So that's the second conclusion.

Conc 2. Timed-Tactical-Angband would require less total encounters in rough inverse proportion to how much more player time is spent per encounter.

This was exactly the issue I was thinking about. Usually only a small fraction of encounters are tactically interesting. Diving increases this fraction, and for many variants it is higher. In Sil's case it is a lot higher, which is fine because it is a fundamentally shorter game.

Another notable feature of Angband is the ability to play it in lots of different ways, and any change needs to consider that.

[Derakon]

The question really is, how many really tactically interesting encounters are there in Angband right now? In one game, I'd wager the number is less than 200 easily. That still sounds like it'd take a long time if you assume that fights would have comparable lengths to those in, say, Final Fantasy Tactics, but you need to keep in mind that FFT is a sloooooow game. Every little thing takes ages to perform. When you prune out all the animations, the game speeds up immensely (compare playing Disgaea 2 with animations turned off). Cut things down to 2D so you spend less time rotating the level and trying to figure out exactly what connects to what and things speed up even more. Ideally the game should spend as little time as possible to properly inform the player of what's going on, and leave the rest of the time to the player figuring out what to do and how to do it. Thinking about the battles is fun, so as long as the game maintains a reasonable pace it doesn't matter much if a single battle is even ten times longer than it would be in Angband.

Some other things:

* You're right that this'd be much more of a tactical RPG system than a straight-up dungeon crawler. It's a different genre and the fights would have to be completely rebalanced, both in scale and in degree of intelligence of the enemies. That said...

* In the case of the player and one other monster exchanging melee blows, this largely devolves into the current Angband mechanics -- both movement and melee attacks occur instantly (no charge time) and have very little recovery time. Of course you could change this if you wanted to. Some interesting possibilities open up if e.g. attacking has a longer recovery time than moving does -- after being attacked you have time to back off or to get partway to charging a more powerful attack.

* At least in the Grandia games I've played, relatively few monsters have Cancel attacks. The player generally does not have to worry about having their turns cancelled, so they're free to bust out the big guns. As for the monsters, while the player can choose to spam their cancellation abilities, those are relatively weak damage-wise. Usually it only makes sense to use one if the timing works out that you can hit an enemy when they're charging for a bigger attack -- and if they start charging that attack shortly after you ended your previous turn, you probably don't have enough time to do that.

* I honestly don't know how well this would work with a solo character. Some of the potential interactions are lost when you only have one character, since you have fewer opportunities to interrupt -- with a four-PC party in Grandia, opportunities to use Cancel attacks effectively happen probably 1-4 times per battle; cut that down to 1 PC and it may be less interesting. Of course you can always make a multi-PC party; just because most roguelikes only have one character doesn't mean it's the only way to do things.

[Antoine]

Derakon, your comments are sounding uncannily like the (non-roguelike) touchscreen game I'm currently working on.

The game features a party of up to 6 PCs, and the action focuses largely on launching timely counters to various monster attacks.

A.