I''d like to talk a bit about different ways of implementing a turn-based game, because an interesting discussion has come up (from the "are turn-based games a dying breed?" thread..). I''d like to focus on the actual turns, how they are taken, and when they are taken. I''ll need to define a few things, and feel free to further specify or detail these definitions, or attack them on their correctness. Definitions are important. 1. A game is turn-based when the time does not advance unless the players take certain actions. There must be a strictly defined upper limit to how much you can do in a single action. Time advancement need not necessarily be synchronous between players, or even between units. An example of this: there are a few turn-based strategy games where the result of your action is calculated as you take the action. This means that time has advanced for the unit taking the action, and the unit receiving it, but units that have not taken actions yet still get to do so. 2. There are two concepts important to turn-based games: "planning" and "execution". In real-time games, this distinction usually is less important, because execution immediately follows planning, but in turn-based the two can be separated. While the meanings of these statements is probably pretty clear, I''ll elaborate: - planning is: making a tentative schedule for a unit, deciding what it should do with its available actions, where it is going to go with the movement rate assigned to it, etc. Whether the exact thing planned happens or not is implementation-dependent, there are reasons why some things may turn out to not happen or be impossible. - execution is: the execution of the plan made in the planning phase, and the calculation of the results of that plan. The style of turn-based play that I see most often consists of taking long turns, with quite a few units, and quite a few actions per unit that can be taken per turn. Usually, planning and execution happen at the same time. Civilisation does this - as soon as you decide to do something, it happens. If you plan to move north a square, you then move north a square immediately. The players tend to go either strictly in turn (one player cannot move until another has finished all of his turn), or go simultaneously (both players can do their moves at the same time, when both have clicked "end turn" the next turn becomes available). The simultaneous turn-taking is very advantageous for turn-based games, in my opinion. Turn-based games are generally slow, and anything that can alleviate the waiting is a Good Thing(tm). However, for the immediate execution style of play, there is a problem. Even though turn-based exists specifically to avoid twitch-gameplay, now you have a case where being the first to act is still advantageous. If both you and your opponent want to move to the same square, and your opponent clicks first, you''ll get the short end of the stick, and your planned move is now down the drain. In my opinion, this should be avoided at all cost in turn-based games. It destroys the point of them, namely having the time you need to make your strategic and tactical planning, without them being influenced by the other players'' reflexes. (note: from here on, I''m always assuming the system to be based around some form of simultaneous turn-taking. If you disagree with this always being a good idea or see other ways of doing things, please post! This is a brainstorm). The problem to solve is then: how to stop the other player''s planning from interfering with yours, even though you''re working on it simultaneously? Actually, I''m cheating since I''ve put the answer up above already, by separating planning and execution. I think this is one possible answer. There are a few advantages to this: 1. While the other player is planning, you do not have to see what he wants to do. If execution was immediate you would see his units move, but now you don''t see anything. Your opponent would see something like a path stippled out for his unit, and a transparent version of it in the location where he wants to end up. 2. Because execution is postponed until the end of planning, you will be contesting eachother''s planning, and not execution. This means that a faster-clicking player has no real advantage over a slower one, besides being able to take regular smoking breaks (ps: kids, don''t smoke, it''s bad for you!) "contention" is still possible though, and there are some other situations that still cause trouble. By contention, I mean that it is still possible that both you and your opponent have chosen for a unit to go to the same tile or location. They can''t both be there, so there has to be some sort of resolution to the conflicting situation. The easiest way to solve it is to assume that the units attack eachother, but I think it''s likely that this is not always the logical or best solution. If you have any ideas on this, do post! One particular other situation that comes up a lot, even in tabletop games, is where you have two sets of cover, between which is a distance less than the maximum move of a unit. Now, even if player 1 has an entire squad of marines with their rifles pointed at the gap between the two pieces of cover, player 2 can still move his entire army across the gap without player 1 ever getting a chance to shoot. The way this is usually solved is by having an "interrupt" action. I think UFO-enemy unknown did this, and I''m sure laser squad did. What happened was: if any units of yours, with enough action points left to do something, suddenly had an enemy unit wandering into their line-of-sight, you''d get an "interrupt" of the action, and you would get the opportunity to let your unit fire at the enemy unit. This solution is neat, and quite elegant for games where planning and execution follow eachother immediately but there is no simultaneous turn-taking. However, for simultaneous turn-taking it is very cumbersome to constantly get interrupted as your opponent blunders into the line-of-sight of a unit with action points left (and if you''ve just started taking your turn, there will be a lot of those units!). Also, when you separate planning and execution, it will get worse. Even though you''ve probably spent quite a while on planning, execution will still get interrupted a lot with more short planning phases as these ''opportunity attacks'' take place. So, finally to the reason I started this thread (are you still awake? I know this is a lot of text )... I feel there is another way to solve the issue, and much more elegantly for the situation I''ve described above. This situation is still the following: Players plan their actions simultaneously, execution only takes place after both players have committed their plan. What I propose is to drastically reduce the amount of action that any single unit can take during a single turn, to the point of having something like "take a single step OR fire a single shot". This mainly has the effect of reducing unexpected situations. The only time you''ll move into the exact same space as an opponent is when you are really very close together. The enemy''s squad will not be able to make it across the gap in a single turn. The advantages of this, as I see them: 1. Less unexpected contention, hence no special additional rules needed to deal with them. Simply attacking eachother when you move into the same space might be enough, other detail is possible but probably not necessary. 2. No special cases needed for opportunity attacks, because you will always get an opportunity to fire in a reasonable setup. 3. Planning time will be much reduced, because there is less to plan. 4. Execution time will be reduced, because the shorter plans will mean there''s less to execute. So, turn-taking will be faster, turns will follow eachother faster, and less special-case rules are needed... Which makes me think: what have I missed, because this just sounds too easy and too good?

I think that Civ, given the scale of the game, actually comes pretty close to "small scale movements" already. Most units can only move a single tile, there are only a few faster ones. I''d hazard a guess the game was so popular exactly because of this - and because you are zoomed out so far, little tactical details such as running from cover don''t really get in the way of player''s perception.

It''s specifically for smaller, squad-based TBS games that it would make a big impact. Things in the vein of Laser Squad, UFO, etc.
With small movements, and a maximum of about 8-20 units per side, you could cycle through all your units giving orders in less than a minute. That means that around every minute-and-a-half you''d get an execute update and another turn.

I''d also allow for planning "ahead" more than one turn. For instance, deciding that your Grenade Launcher marine should move to the top of the next building, and setting that path. You expect that nothing will occur before the marine reaches his destination, so the "default" movement for each following turn would be to continue the path and this would be set automatically, without needing confirmation, but if you feel like interfering you still get that chance. This is similar to "interrupt gameplay", but it doesn''t disrupt the normal turn flow.

The AI idea would also still work here, and in the same fashion. The AI could decide on the default actions of units at the start of each turn, and somewhat according to your longer-term planning. Each turn, you have the option of just clicking "accept planning" and letting the AI do the rest, or tweaking the actions of each unit, changing long-term planning, etc...

This way, I think it might be able to handle much larger numbers of units without slowing down the gameplay as much as under large-step turn systems

I am a wargamer and have played many different type of board-, war- and tabletop games. Besides that I also have played computer wargames. The problem of turn based games is not new, some of it was even dissected by James Dunnigan - How To Make Wargames (you can find the online transcription of the book at www.grognard.com).

By definition, a turn in a game is a timeslice during which one player may perform one or more actions. Historically games started with turn system as simple as you go I go:
Turn
Player A
Phase 1 - Move
Phase 2 - Fire
Player B
Phase 1 - Move
Phase 2 - Fire
Already in the 70s wargame designers were creating monster games and were hitting the main limit of that system: slowness. Turn systems were then developped and more elaborate (some example below)


Of course those turn systems are only for tactical games. When you get to grand tactical, operational and grand operational types of games you add new actions inside your turn system like diplomacy, production or reinforcements. The main point that has been made during all those years is that the turn system deeply reflects what you want to emphasize in your game (the role of infantry, the importance of planning reinforcements, the restrictive effect of diplomacy, ...)

By this first presentation of turn systems I wanted to answer two points in your definitions:
1- that the definition of a game turn has evolved a lot during the gaming history. I think your turn definition may be closer to that of a phase inside a turn.
2- the concepts you developped between a planning phase and an execution phase may not be true for all the type of wargames (simple operational games like Victory from Columbia Games or Grand Tactical games like NATO from Victory Games may not be translated on computer using the system you propose without changing the focus given by their turn system). These concepts you develop show that you plan to make a turn based game focusing on the importance of planning (as the rest of your post develops on why you prefer to decompose a turn between these two phase planning/execution)

Now I will adress the faults you found in a turn based system and show the solutions that board and wargame designers used:
1- The turn for each player is too long making the game slow with long pauses
The turn begins to be slow when the number of playing elements is over 15. It is all the more slower as each playing unit may have a lot of actions to perform.
Either you design mass combat systems, then the rules are streamlined with very little number of actions to perform. Only the stats of the playing units vary and they are few (move - attack - defense).
Or you reduce the waiting time between players. A lot of games use alternate move sequences: Player A moves one unit, then player B moves one unit and so on until everything is moved, then fire is resolved simultaneously and casualties are removed after all fire has been resolved.
2- Being the first in a game may give an edge
Initiative rolls at the start of the games are used to break the prediction of who may go first each turn. This gives players the need to plan ahead since they are not sure to always be the first to start.
Be aware that in alternate move systems, this does not have much impact. In simultaneous fire systems, this does not have much impact too since all the casualties are removed after all firing is resolved.
3- The case of opportunity fire
Opportunity fire is used for tactical games: it reflects the use of squad tactics when conquering land. One quad moves while the other does not to cover any unwanted counter attack. You may even go further and add suppressive fire: one squad moves while the other fires at known enemy positions to avoid opportunity fire on the first squad. (Anybody knows of Advanced Squad Leader by example ?)
Now in your system, you say that opportunity fire worsen the time for one turn since each player must plan before resolving opportunity fire. I must disagree: a normal wargamer always plans for opportunity fire before moving(Age of Rifles did implement that one correctly). You do not need to add planning phase before each opportunity fires. Planning is done first, then when you resolve the actions all fire is done accordingly to orders given to units during the planning phase (move or fire).

Now your solution does not answer the problem of making turn based games for computers: I think it answers the creation of a computer turn based tactical game. The arguments like less contention, no opportunity fire needed, execution time reduced are valid for your system. I do not agree however with your argument that the planning time is reduced: the planning time is dependant on the player. From my experience some player take a very long time to decide themselves on what to do. The only solution found by designers is to impose a time limit: when that time is spent, actions take place whether the player decided himself or not.

As to what make a computer turn based game slow, I would say:
- lack of clear and unambiguous informations: you take time to understand what is your tactical situation.
- Number of units too high: you do not end a turn of 5 min with 2000 units in your army
- Number of actions per unit too high: you should never have more than 2 special actions per units unless you have only 5 units in your game.
- undecided player (see above)

Hope that helps,
Ghostly yours,
Red.

quote:Original post by Red Ghost
Now in your system, you say that opportunity fire worsen the time for one turn since each player must plan before resolving opportunity fire. I must disagree: a normal wargamer always plans for opportunity fire before moving(Age of Rifles did implement that one correctly). You do not need to add planning phase before each opportunity fires. Planning is done first, then when you resolve the actions all fire is done accordingly to orders given to units during the planning phase (move or fire).

But you still need to do the opportunity fire. This is less of a problem if the scope is very limited - for instance, if you can only do opportunity fire with the currently equipped weapon and the fire mode enabled during planning. However, if you want a "reaction" type of gameplay, then the player will need to make certain choices when opportunity knocks, and this will pause the "execution phase" in my above design.
Note that I do realise that in differently designed systems (movement/execution style, and interspersed turn taking as you described) this "pause" is much less of an issue, it is simply part of the normal turn-taking process.

quote:Original post by Red Ghost
Now your solution does not answer the problem of making turn based games for computers: I think it answers the creation of a computer turn based tactical game.

You could be right, there are easier or more involved solutions for other types of games - but my mind tends to focus on the small-scale turn-based games. I have rarely played very large-scale wargames, where you control thousands upon thousands of units. The biggest I''ve played is Space Marine by Games Workshop, but there they kindof cheated by treating entire squads of 5 small units as a single unit anyway.

quote:Original post by Red Ghost
I do not agree however with your argument that the planning time is reduced: the planning time is dependant on the player. From my experience some player take a very long time to decide themselves on what to do. The only solution found by designers is to impose a time limit: when that time is spent, actions take place whether the player decided himself or not.

Yes, and no. The player will always be slow, regardless of the system you choose, BUT, if you create a buffering method by simultaneous planning, you will ALWAYS shorten the time you have to wait for your opponent to finish, namely, you''ll take off the time that you''ve spent planning your turn. This is an improvement, though depending on how much the players differ in decision speed, it could be a slow one.
The imposed time limit can still be used in my suggested case, and is probably a good idea. One way of doing it is to not have it per-turn, but rather (mostly) for the whole game, like it is done in chess.

Well here are some of my thoughts, since I have been designing a few turn based stratergy games lately.

I think the best model for multiplayer computer use with stratergy games is a similatnious planning/computer ai defensive model.

Active Offense/Static Defense

Each turn has two phases, movement/combat resolution.

The idea here is that the player conrols the offense of his units by laying out a move for each unit or targeting a specific enemy.

But units not specifically moved or targeting are static units that are "defending". Defending units will automatically engage enemies that enter their radius of control if such an engagment is calculated to turn out favorably.

This kind of automatic defense is used to offset the similtanious planning phase. For instance imagine a game of chess with both people planning their moves in advance. Without any static defense it imediatly becomes obvious that the whole game would be determined by who wins the initiative on vital turns.

But with a static defense it imediatly becomes obvious that each player will have to make good moves to stand any chance of "winning".

Ok my other thoughts was related to time. In my model I was planning on giving each player an amount of time per peice for the average number of peices owned by players. This lets the game move quickly in the early phases and then slow down when more peices became operational.

Also I like this model because it gives the smaller guys an edge on strategical planning while punishing the bigger guys a little, hopefully this helps to maintain the balance of power a little as it is often just the guy that gets big first that wins these sort of games.

Cheers,

Jai Shaw


--------------------------
Dreamstars 3: A space combat game with depth?!
http://icarusindie.com/Dreamstars

I''ve been working at a single player turn basied game that is somewhat X-Com like...difference being that players control a small squad of rescue personnel out to save people form burning buildings and other such dangerious situations.

I''m working at implamenting a "turn que"...that is, the player can only move one of thier team mates at a time...all thier planning and possable actions that can be taken revolve around the single unit rather then the whole squad...this shortens the turns by haveing the player focus only on one unit...it should help with "planning" because all other AI game elements take turns in the same manner...the player doesn''t have to waste extensive amounts of time trying to figure out what the AI has done during its turns.

The "que" works by makeing a list of all active units...then sorting it basied on a "speed" attribute...the list is then processed one unit at a time, until all units have been processed...then the list is recompiled, sorted, and run through again...certain actions can reduce the "speed" attributes of units (as well as increase them)...so for the next "round" (list compiled and sorted) units can be at different locations within the que.

quote:
But you still need to do the opportunity fire. This is less of a problem if the scope is very limited ...
(...)
my mind tends to focus on the small-scale turn-based games ...

I would like some precision here. When reading your posts, I have the impression that you are looking to make an infantry tactical turn based game (whatever era that may be) with as little rule as possible. To me you want to put an emphasis in your game on planning since you will leave no control of the action phase to the player (it is automatically resolved using a few rules). From that I infer that your end-user is a casual gamer and/or a beginner in the wargaming hobby. (From what you tell of your tabletop gaming experience, I have the feeling that you were influenced by Warhammer 40K from Games Workshop - please correct me if I am wrong). Could you tell me if your solution is designed for a precise kind of turn based game ?

I stress that out because your post started with :

quote:
I''d like to talk a bit about different ways of implementing a turn-based game ... (...)
I''d like to focus on the actual turns, how they are taken, and when they are taken. (...)

To my mind the solution you found out seems more limited in scope than what your starting post seemed to imply.

Regarding turn based computer games, you can have as much level of "realism" or complexity as you wish: just look at Steel Panthers, the Star series or Age of Rifles. The systems work while presenting the player with a simple GUI to act with.
You can see another answer to turn based game at that following web page www.silicmdr.com/index.htm (look for Solar Vengeance) where the designer created a complex turn based game quick to play and simple to learn. Here his emphasis is not on planning before action but on managing resources under a time pressure.
Electronic Arts in its turn based game Space Hulk (adapted from a defunct Games Workshop game of the same name) had another approach. The game was played by impulses of 1 second during which all events happened (move of the aliens, reaction fire, explosions); the player could give orders to each of his 5 marines during that period but the orders took time to process and he could not coordinate his squad (the pace was too high and each second was valuable against the Genestealers). But for each second the player was letting the game play, he was increasing a reflexion time counter by one second. Whenever the player paused the game, he had a pause time equal to the amount of reflexion time gained to 1)observe what happened and to 2)give full orders to all his marines (like to coordinate an attack while guarding his left flank).

Now back to your argumentation:
Regarding the opportunity fire, I think we may not have the same vision of the player. Of course some player will want to react to opportunity fire. However the consequence of opportunity fire is either to stop your unit ("realistic" when compared to soldier training), let the unit continue its order (just mosquitoes guys ! keep going ! ), or retreat (even flee if the guys lose their nerves - "realistic" when compared to battle experience). The time frame when using opportunity fire does not allow much more reaction id est the designer can choose what type of reaction will result from opportunity fire considering his emphasis ("realism", fun or other ...).
When looking at gamers confronted to opportunity fire, they often tend to plan before moving and to make contingency plans if their squads ever receive opportunity fire (and I am not speaking about the hardcore wargamer who will find that giving time to react to opportunity fire is not realistic !). To me it is rare that a player wish to stop action to ponder what means an opportunity fire in terms of tactics. He will rather see the result and act upon at the next turn. Do the players you are confronted to have a different reaction ?

Regarding buffering orders, I understand you act upon the fact that all players have an individual PC and need not wait for the end of one player planning to plot their moves. I think it is again a design choice between letting each player see what the other is doing and react to it (Heroes of Might and Magic by example) or to hide each opponent moves (in that latter case your solution is perfect since you should not wait). In the first case, the fact that you look at your opponent doing his moves is not necessarily a loss of time (some wargamers find it instructive since they use that time to plan their next turn using what information they see).

Hope that helps.
Ghostly yours,
Red.

I haven''t read anyone else''s post here, so sorry if cover something someone eles has written:

quote:Original post by MadKeith V
I feel there is another way to solve the issue, and much more elegantly for the situation I''ve described above. This situation is still the following: Players plan their actions simultaneously, execution only takes place after both players have committed their plan.

This is exactly how I''m designing my game. It will be a hybrid turn and RT system. The game is broken into turns which consist of each player making plans for all of their units. In my system, I want very detailed and "chainable" system of orders, with some limited decision making algorithms built into the GUI order system (for example, IF fired upon, THEN move to position). Once all units have been ordered from both sides, the execution takes places in real time. This is the all around best compromise and I believe offers the best of both worlds.

Once the real time phase begins, it takes place within a set amount of time. For example, say 15 seconds of real time, that represents 1 minute of game scale time. Units will accomplish what could realistically be accomplished within that time frame. For example, if a unit was moving at 60mph, it could travel in game scale...one mile.

After the 15 seconds (or however long you wish to make it) is over, it goes back to another turn phase. In my system it will be possible to alter the flow of events, but it won''t be guaranteed to go through, and there will be a slight delay for the unit to react in mid-course. This will allow for some slight tweaks in case your opponent did something unexpected.

The other advantage to planning your moves is it requires you to really think in advance what is going to happen. I believe that most RTS games the way they are designed are more reactive games...where you have your plans, but are forced to react to your opponents actions because of the RT nature of manipulating your units. When you have a turn based planning system, it makes you more proactive, and makes you think ahead of time more....trying to guess what your opponent will do, and not requiring your attention to be focused on the manipulation of your units.