Simultaneous turns on tactics game

[pakoito]

This is not a full response to your questions as I'm trying to figure your concerns out myself. I'll try to draw-test stuff myself tomorrow. Probably single-action or single-char is the answer, but I wanted to know if multiturn works.

Spells: AoE damage. Take fireball. It explodes in a circle on contact on when it reaches target. Zoning again, because the second player doesn't know where it's going to land. Lighting strike may reach its target instantly and jump to any unit within 2 spaces.

Melee: this one is quite a conundrum. I guess 1-square range cone/whirlwind attack is what makes more sense to be able to hit more people. Threat areas as a way of autoattacking maybe?

About ranged attacks: projectiles are shot from the player's hex and move with the game turns. If you target an hex, the arrow would move from your position to the target at a certain pace (1-3 hex per turn). This allows for zoning more than damage, and the reason why I'm using hexes instead of squares. Two arrows moving through two lines of the board take more space than it looks. Also archers would have long range threat zone to disrupt spellcasting.

Drawing time! This is my first prototype. Every dot is a character. Every handrawn icon is an "arrow". In this ruleset you moved like chess only one piece per action both players simultaneously, and you have to declare 5 actions ahead. No collision rule yet. Arrows move before the player's action at the pace of 1 hex/turn since fired. One hit kills. In this prototype keeping track of them was complicated but we could only simulate 4-5 turns.

This is turn 3. Blue arrows are coming down diagonally, the yellow one is continuing to the left in a line following the botton of its current hex. What are yellow's next 5 moves? The leftmost toon has to waltz and guess where rightmost blue will move, but yellow's first turn has to be used to be centermost toon out of the way, so he's at a disadvantage.

 

[pakoito]

I was experimenting for a while and came up with another completely new system based on the waging gimmicks. I think it can also keep it multi-turn and deal with collision.

We have a new per-player stat: gauge meter (or coins, or pills, or mana, it's just a number). Every character adds an amount of gauge to the meter at the beginning of the turn. Maybe the gauge already has a base value when the game starts.

Now, the actions available (with arbitrary costs)
- Move: costs 2 gauges.
- Spell/skill: costs X gauges.
- Counter: special skill.

Turn sequence:
- Each character adds its gauge regen value to the pool.
- Each character gets 2 personal temporary gauges.
- The player chooses an action for every character. The cost can be taken from the temporary and the meter.
- If the player wants, he can spend gauges from the global meter to make its toons do extra actions.
- Actions are resolved simultaneously, but respecting initiatives. Missiles like arrows or spells are considered of initiative 0 so they are moved at the beginning of the action.
- In case of collision or counter, the highest bidder gets to do the action. In case of draw in the bids, initiative wins. Both players spend their gauges and their skills are put on cooldown, but the loser cannot take more actions this turn and loses any other gauges he has spent for further actions. Any other actions he could have done are not put on cooldown.
- At the end of the turn, all unspent temporary gauges are taken away.

Extras:
- Extra gauge can be paid for unsafe skills to avoid being countered.
- Extra gauge can be paid for any skill for extra effects, if the description says so (i.e. extra range, extra duration, extra damage, in-action delay)


Action examples (arbitrary numbers):
Move
Cost 2
Range 1
Initiative Depends on char, range 500-600.
Cooldown 0
Unsafe False
Movespeed Instant
Type Physical
Effect: Moves character one hex
Enhances: 1 gauge for +500 initiative delay, 1 gauge -20 initiative delay

Fireball
Cost 3
Range 1-4
Initiative 345
Cooldown 2
Unsafe True
Movespeed 2
Type Magic
Effect: 3 damages in radius 1 from target. Explodes if enters an occupied hex.
Enhances: 1 gauge for +500 initiative delay, 1 gauge for max range +1, 2 gauges for damage +1

Grappling hook
Cost 4
Range 2-3
Initiative 435
Cooldown 8
Unsafe True
Movespeed Instant
Type Physical
Effect: Moves enemy to your adjacent hex. Cancel the rest of his actions this turn.
Enhances: 1 gauge for +500 initiative delay, 3 gauge for max range +1

Swing
Cost 2
Range 1
Initiative 820
Cooldown 0
Unsafe False
Movespeed Instant
Type Physical
Effect: Deals 2 damage to all enemies in arc
Enhances: 2 gauge for -300 initiative speedup, 3 gauge for +1 damage

Arrow
Cost 2
Range 10
Initiative 830
Cooldown 1
Unsafe False
Movespeed 1
Type Physical
Effect: Deals 2 damage the first enemy it crosses with
Enhances: 2 gauge for -500 initiative speedup, 3 gauge for +1 damage

Break Concentration
Cost N
Range 1 (Swordman), 1-2 (Spearman), 2-3 (Archer)
Initiative 10
Cooldown 0
Unsafe False
Movespeed Instant
Type Counter
Effect: Stops any unsafe actions done within range with cost N*1.5 rounded down until the end of turn. Enhancements do not count towards cost.

Counterspell
Cost N
Range 1-5
Initiative 11
Cooldown 2
Unsafe False
Movespeed Instant
Type Counter
Effect: Stops the next unsafe magic actions done within range with cost N until the end of turn. Enhancements do not count towards cost.
Enhace: Pay M to stop M additional unsafe magic actions.

 

[tuluse]

You need to have branching plans for turns. Frozen Synapse doesn't have to worry about units standing on top of each other not because of instant death, but because the units react to what's happening. If two enemy units see each other they start shooting instead of continuing to walk.

 

[pakoito]

You need to have branching plans for turns. Frozen Synapse doesn't have to worry about units standing on top of each other not because of instant death, but because the units react to what's happening. If two enemy units see each other they start shooting instead of continuing to walk.

Have you read the post above you about the new system? Seems to deal well with it, tho it deviates from the original idea.

And with instadeath I meant that if two units see each other, a single automatic shot suffices to kill. In this system several attacks would be needed and there's no fog of war or LoS mecahnic, but a timing one.

 

[baturinsky]

Does moving to other spot first thing in phase make you effectively invincible to most attacks?

 

[pakoito]

Does moving to other spot first thing in phase make you effectively invincible to most attacks?

Mechanisms to counter this: I have to change move to be unsafe so it can be punished by high initiative counter. Second, projectiles are used for zoning, you can deny or make several movements costly if you place them well. Third, two points gives you an extra move action, and in that action projectiles will move too. The board is not specially big (I'm leaning towards 10x7) so a move 2 projective would cover the biggest line in 5 actions. Fourth, I can make melee moves like swing act in a cone, or similar, or spend gauge to make them into whirlwind attacks.

 

[baturinsky]

Still sounds kind of like a brawl in a pitch black room. You sure 1x4 board (front and back rows for both sides) would not be better?

 

[pakoito]

It's iterating, so why not. I'll try to get a friend to play with me this weekend, and if any codexer has enough patience to play on a whiteboard or a program like Vessel, just contact me.

I'm already coding general stuff like movement and user input, etc so I won't be sitting idle.

 

[Galdred]

I thought about going simultaneous turn based too at some point, for a melee heavy Tactical RPG. The problem I had was that melee makes it much more cumbersome to plan ahead than the firefights in Frozen Synapse or Combat Mission (which I already find much less fluid than IGoUGo like Jagged Alliance or Steel Panthers). It is very hard not to make such a system very complicated, thus slow to use (like in Knights of Legends), because you either have to force very short turns (1 move/turn or something like that), or very high uncertainty (5 turns ahead with melee brawls will end up as a complete mess more often than not), and end up with a "light programming language for player actions" or something really too cahotic, which is why I settled for a classical turn based structure in the end. I would definitely go simultaneous turn based for a strategy game (because of the higher level of abstraction, allowing you much larger options to shape the world rules), but for a skirmish level engagement focusing on melee, I just could not come up with a satisfying system.

That said, in order to make it simultaneous (much better for MP and mindgames indeed), I too came to the conclusion that the gauge/bid system would work best (I like initiative + some randomness too, but not as much).

We can vassal it someday if you are still up to it.

 

[pakoito]

I was able to test the system with a friend and numbers up or down the system works. I'll be glad to play with you some day this week.

As for the programming part, this is how far I am:

Alternate per-player turn sequence with one action per creature is what's currently in. I can even make it one action per player easily too. Actions are resolved simultaneously too.

The biggest problem now is the skills as the scripting system even though it works has lots of problems with implementing skill interactions. It'd be better to hardcode them to be honest.

 

[Haba]

Why you steal mah ideas! God damnit!

I've been toying with the same concept. Will post more once I've the time. Would read again,

 

[Severian Silk]

I played Laser Squad Nemesis but couldn't really get into it. I had trouble remembering all my moves. I guess my short term memory is bad.

 

[Severian Silk]

I've played Laser Squad Nemesis before and it didn't really click with me because it's hard to remember movement orders and stuff. Maybe my memory isn't the greatest.

 

[pakoito]

Can't work on this ATM but would love to develop the idea a bit. My current in-brain iteration moved away from simultaneous to per-team, and instead of having a gazillion scriptable skills use a system similar to Bonfire's, in which you get a different spell for yourself, floor, allies and enemies. You get a discrete set of skills variations and probably just hardcode them.

 

[Ranselknulf]

I skimmed this thread.

So you are doing simultaneous turns but each unit gets a certain amount of "action points" per turn? That would make more sense than 5 turns at one.

 

[pakoito]

I skimmed this thread.

So you are doing simultaneous turns but each unit gets a certain amount of "action points" per turn? That would make more sense than 5 turns at one.

Doesn't fix "turn break" See how Frozen Endzone dealt with it, there's only a discrete number of results to turn-break, and is done per-character. If they get blocked they lose the rest of their turn. Very nasty in a context of a complex multi-skill combo-based game.

 

[Ranselknulf]

What do you mean by "turn break"?

My assumption is you mean the time between turns where people formulate their strategies, actions, and moves. I don't see how giving people 5 turns with a specific amount of "action points" is any different than just giving players one turn with more "action points"

Lets say each turn your unit got 3 action points and you did 5 turns in a row. That would be 15 action points. Then you iterate through each action point for both players to get results.

How is that any different than having one turn with 15 action points. You would still end up at the same result. You aren't giving the player a "turn break" to change strategy between those 15 action points so calling it 5 turns really is meaningless. It's just one turn with 15 action points. You could label every 3 moves a turn but in my mind a "turn" means you are allowed to change/adjust your strategy.

 

[pakoito]

Simultanous turns implies a planning phase and an execution phase. A "turn break" happens when one of your toons cannot perform the actions he was assigned because there is some interference (he gets stunned, the path he was walking through gets blocked, his pet dies...) so the statespace created is too big to handle for the plan anything meaningful. It would be like trying to play chess 5 movements ahead each player, even resolving them sequentially requires a very specific "resolution ruleset" in place.

 

[Ranselknulf]

My entire point was it wasn't 5 turns but one turn.

but in regards to how to deal with dynamically changing events, i'd say it would depend on just what the event was. Like stun you could just drop the last "action" in the units "action queue" and that would make them lose an action because of stun.

If you want to minimize the conflict for long chains of actions then allow players to interact with their "action queue" more often. (ie less actions per turn)

 

[pakoito]

Even two actions per turn proved as too much, if you see the gif above my current iteration was limited to one action per turn. Abilities cost mana from a common resource pool and had a kewldown because I like my game tropes.

 

[Ranselknulf]

I haven't tried to code a game before but I don't see any reason why more than one action per turn would eat up a huge amount of processing power.

Make your action queue. Make a function to interact with the array of actions. If a unit gets stunned or there is a conflict "like two units occupying the same square" make a special function to deal with that situation.

There are lots of simple games out there that do this like final fantasy tactics etc.


I guess I should clarify also. Computers run on such a different time scale than people so even if all the actions aren't simultaneous in "reality" they will appear that way to us. The computer can set up all the actions one at a time then make a short video where everything appears to happen at once.

 

[pakoito]

I haven't tried to code a game before but I don't see any reason why more than one action per turn would eat up a huge amount of processing power.

Nobody said it does. It doesn't. State space is not a techie term but a design one, it's all the possible actions and ramifications. I'm not talking about AI but human behavior.

Let's say we both play chess, except instead of alternating turns I write my next 5 moves, you do the same, then we resolve them sequentially. Do you really think I can plan any short or mid-term strategy? That I can think of all the possible enemy moves in the precise order? that is the biggest problem. Previously in this thread we established that no, or very probably not. The second biggest issue is "sequence break", where I plan a movement that requires that a previous step has happened (i.e. occupying the space of a piece I move 2 turns before). You have to define a clear set of rules for resolution of those collisions, the most common being "piece's action queue gets cleared and loses it's turn". With a 5-moves-ahead you may find that right from the first move you just sent your pieces to die in a way you couldn't really foresee. It's not tactical, it's a clusterfuck, you can't make a sane analysis and come up with an optimal solution, just vague approximations.

As this design ideas do not exist in the void I saw what others did, and the closest example is Frozen Synapse/Frozen Endzone. In there "sequence break" is solved by substituting the action with a small set of engagement rules. This is fine because their possibilities for a turn are very few, and common for everyone (move, pass, wait) and their engagement rules terminal (blocked until end of turn or death in shootout). What my game wants to do is too complex to use that same system, I can't create a 10-second round where skills are supposed to interact between them in timing (spiderweb into oil into fireball, D&D's oldest combo, or stun into hookshot for Dota fans).

As for the techie part, weighted minimax algorithm with pruning does the trick for any simple ruleset, but competently playing this game, same as Civilization or Dominions, would be impossible for modern AIs. You cite Final Fantasy Tactics as if it was simultaneous-turn. It isn't. No turn-based tactics game in the market is, they either do a per-character or per-player approach to turns, period. Hence why this thread flexing the design muscle for some forumers.

 

[Ranselknulf]

My mistake, I was working under the assumption you were talking about a turn based game.

If you make it turn based you can still have all those complex interactions after user input.

If it's turn based it is entirely possible to have a 10 move chain of abilities that gets interrupted and substituted. For real time you would need something completely different and I'm not sure how to help you. I haven't even considered the idea.

 

[pakoito]

Well, it's right there on the title lol

 

[Ranselknulf]

I took "Simultaneous Turns" to mean both players input their actions on a "turn" then you watch all the actions play out at the same time.

Otherwise you may as well call it real time tactics or "RTS" game. It would be easier to understand.