Thoughts on Multiplayer from the developers...

from Stardock Forums

I was thinking (happens once in a great while) about how one could do multiplayer in turn based strategy game.

Then it hit me, why not have the game exist not on any of the player's computer but instead reside on the server as part of a database.  This would combine the best features of play by email and RTS together. Let me explain:

Player Joe starts a new game. The game is then added to the Metaverse as an active game. Joe gets a special "master" code for his game. Other players can then join in at any time. Player Joe can set it so that only people who know the guest code can join in or any player can join in. Each turn updates the data on the master server so the data is persistent on the server.  Players who join in at the start get a guest code.

Dropped connection? No problem, just reconnect and continue where you left off, an AI player can take over while you reconnect. Then just enter your guest code and continue.

Long game? No problem, just agree to play X number of years and resume a week later.  If that next week one of the guys can't make it, you can continue on without him (unless he's the game creator i.e. the "dungeon master" so to speak) -- obviously it would be polite to get the missing guy's permission to do that.

And because it's handled by a central server, you could have smart turns. This would be something in between simutenous (sp) turns and individual turns. Smart turns would be such that if two human players can't possibly interact in a given turn with their units, they can move at the same time. Otherwise, any units in areas that conflict would have to go in player order.

And by having the game data on the Metaverse rather than one of the player machines, other players can optionally be able to check out the stats on the game in real time. Almost like box scores in baseball.

But the key thing here is having the persistent server handling it all. It reduces the bandwidth needed and eliminates the whole "disconnect" problem since players can jump back in later on or you can have other players jump in or just exit and continue later with fresh players to pick up where you left off.

These are just some thoughts and there won't be any multiplayer system in GalCiv 1.x and if there is a 2.x it probably wouldn't be in that initially either. But it seems, on paper anyway, that this kind of system is workable.

1,767 views 27 replies
Reply #1 from Stardock Forums Top
If you could do that!It would attract a whole new (and I believe very large) strata in the gaming community. While I'm absorbed in GC at the moment I miss multiplayer, there's just no substitute for interactive gaming.

I know the long term goal of GC is to be able to emmulate any players style on the Metaverse and maybe some time off in the future that could be possible. But what you suggest seems quite viable and would be so warmly accepeted . . .

What I wouldn't give to challenge some of these players in a one on one . . . :d .
Reply #2 from Stardock Forums Top
multiplayer in turn based
End of quote


Certainly viable - the other option is "simultaneous" turns where each player updates the system (wherever it may live) either directly or via net/email/sneakernet or whatever. There is an interesting email (and other format) based code versioning system called "code co-op" iirc - useful for both remote developers and to get the grey cells moving on alternative approaches.

Alex


~SDC~
Fences - Organize your desktop icons today!
Reply #3 from Stardock Forums Top
As a thought - the problem with a "metaverse" based option is that it is totally dependent on access - it would be better if there was an option to multiplay on a LAN or elsewhere....

Alex


~SDC~
Reply #4 from Stardock Forums Top
Frogboy,

A few thoughts here. This system, while an improvement over the current methods of "true" multiplayer and PBEM, still doesn't solve the basic problem - turn-based multiplayer games would take a very long time. GalCiv has streamlined the gameplay and interface to allow people to take days to play a game (me) or minutes (EBZ). But you still have to look at the lowest common denominator when you're joining players together, and some people are simply going to micromanage and play it slow. Slow = waiting = boring.

I realize you're not looking for a vote here, or this would be on the "poll" window. If it's technologically feasible, that's great; as a game developer, you obviously know more about this sort of thing than I ever will. But if you're looking for a player's perspective, for a multiplayer game I look for first-person shooters, and consider turn-based only for single-player.

I don't mean to widen your scope here, but have you considered a real-time option? In the initial game setup, pick a speed at which "game months" pass, and have EVERYBODY go at the same time? Coupled with your idea of a central host, this might be a whole new dimension for GalCiv. If someone is momentarily disconnected, have the AI run them for a bit based on their strategies (the future Metaverse IS collecting them all, right?) Individual players could micromanage all they want, but would have to keep an eye on what else is happening at the same time or pay the price...

Lightsaber in hand, I now calmly await the onslaught.
Reply #5 from Stardock Forums Top
Odi Wan, have you ever played golf on the internet like TW2003 and the like? Thats pretty popular and its essencially turn based. All you need is a clock and time-out options.

Come to think of it Golf in reality is essencially turn based too . . . mabye my analergy is to close?
Reply #6 from Stardock Forums Top
My problems with multiplayer.

A: The quality of most players is not up to my standards. Too much cheating, cheat codes, oops my internet went off, thus you didn't really win, ie the jerk factor

B: Too much time to get through a tbs game. I'm a quick player without much time (hour a day maybe) to play. Average player is much slower but has more time. Not many compatible with me.

And because it's handled by a central server, you could have smart turns. This would be something in between simutenous (sp) turns and individual turns. Smart turns would be such that if two human players can't possibly interact in a given turn with their units, they can move at the same time. Otherwise, any units in areas that conflict would have to go in player order.
End of quote


I see what you're getting at. But it would quickly end up going mostly in order anyway. For example

6 players a, b, c, d, e, and f with turns in that order.

A can't meet f
b can't meet c
d cant meet a.

f can neet e so f must go after e rather than with A.
e can meet everyone (he got the so he must go after a,b,c,d which relegates him to 5th and thus f to last
B and c could go at the same time, but after a.
D can meet b and c so must go after them.

Thus,
a
b and c
d
e
f.

The smart turn would be alot of extra coding for not all that much time gain except at the very beginning. And you'd still have to wait for the slow player to go to the next turn. Nice Idea though. But I wouldn't see that the extra coding would be worth it.


Simultaneous turns only work in games where you can only go to planets rather than a game like galciv where you can go all over the board.

It would probably be better to use the hotseat system or just allow travel to planets with lag times and such like moo2 and go with simultaneous turns

Reply #7 from Stardock Forums Top
Simultaneous turns only work in games where you can only go to planets rather than a game like galciv where you can go all over the board.
End of quote


I could just picture two people trying to attack each other, there ships move all over the place trying to get to each other and never meet hehe.
Reply #8 from Stardock Forums Top
I could just picture two people trying to attack each other, there ships move all over the place trying to get to each other and never meet hehe.
End of quote


This problem did in fact happen in Lords of the Rhelm. I could move my 50 strong peasant armies all over whilst my enemy simultaneously tried to catch em with 500 crack units all the while my peasants scorched the you know what out of my enemies shires.
Reply #9 from Stardock Forums Top
Actually thats an interesting point regarding attacking . . . I always assumed if you clicked on a ship your attacker would track them. Noticed the other day that it ain't necessarily so . . . by the time my ship got to the place where the ship was I clicked on it was long gone. Mabye thats just a default setting thing I need to switch off though, I recall seeing something about it but forget . . .

PontiusBruinPilot sorry I lost the plot when
'B and c could go at it the same time, but after a.' :notsure: .
But I get the gist of it (I think) . . .
Reply #10 from Stardock Forums Top
Hmm, brain-storming time...

Sim turns - you might have a problem with people picking up information about hidden ships before they're supposed to.

e.g. two players, A and B (move in that order), all
their ships have move rate 4 and sensor range 4

A's ships are 1,2,3
B's ships are 4,5,6

a = limit of A's sensor range
b = limit of B's sensor range


(this is just one-dimensional for now)


..1...2..3..ba..4..5...6


Right now, A and B can't see any of each other's ships.

With simultaneous turns, A will be able to move all his ships right away - 1 & 2 aren't near anything, 3 & 4 could fight, but it's A's turn to move first, so A moves first, with no warning about B's ship nearby.

On the other hand, B can move 5 and 6 right at the beginning of the turn, but he can't move 4, cause of the potential conflict with 3. So right away, an alarm goes off in his head. He knows someone's out there. Once A's moved, B might be able to see him or not - but even if ship 3 stays right where it is, B's got extra information that his sensors shouldn't have let him get.

Oh, and cloaking ships would seriously **** this up. If
B can't move ship 6, and he's got lots of sensors there, knows no enemy ships are visible, then he knows a cloaked ship's around, and scrolling through the ships that can't move would give him a good guess what region the cloaked ship's in.

Randomly saying who goes first would make it a bit fairer, but still confuse people. If a ship's hidden, sometimes you get the warning, sometimes you don't. Messy.


Real-time? You could have that, with the ships pinging for attention when they reach their auto-pilot destination, and a research "message received" icon instead of it taking over your screen, so if you're in the middle of a battle you won't get messed up.

Actually, a real-time thing brings up the option of real time combat, where the attack/defense turns actually take time. BANG, pause, BANG, pause. That'd give a chance to retreat, or call for reinforcements. Of course, if we let people retreat, that'll just give a HUGE advantage to people with speed bonuses, since they can run from losing fights, and chase enemies if they run away, and keep shooting them in the back. Perhaps whoever retreats gets to be shot in the back a few times? Hmm. I don't think this'll become a RTS any time soon.

That research thing just reminds me - how about a research queue? Something dynamic, so once you add "Comm Theory" to the queue, "Universal Translator" immediately pops up as an option. People could set up their favourite queue at the beginning of the game, and then just tweak it based on the situation.


I don't usually like fixed time limits in MP games, they make the game either too slow at the beginning or too rushed at the end. How about a time limit that changed based on the number of planets and ships you have?

If people are gonna be playing quick multi games, the smaller maps are going to get a lot more gametime than the gigantic maps. Perhaps pay some more attention to issues common on the small maps (e.g. how long it takes to get very good /evil morality without using the re-colonize cheese)

Pre-defined maps? Like for tournaments, a set of balanced maps for all five races. People can see all the planets off-line, so they can plan different strats for different maps, and no-one can complain about a bad start position.

Jerk factor - This could be a major problem, especially considering how a loss can be a real ball-and-chain on your metaverse score. We've already got player records here, just add something like they've got on yahoo games - no. of games played, no. of games dropped. Everyone will have dropouts once in a while, but you can tell someone's playing silly buggers if half their games get cut off. (That'll only work for net games, not LAN games though)

LAN games - jerk detection and watching the game status through the net won't work, but you should still be able to set up a game on a LAN. Submitting it to the metaverse? Well, it could work, with score adjustments for opponent skill (using their player metaverse scores as a very rough guess of skill). Then again, if one person gets a win, and four others get a loss with a next-to-nothing score, the four would probably smash the winner to make sure he didn't post the game. :-)

[Message Edited]
Reply #11 from Stardock Forums Top
Interesting and fresh approach. I think this could surely work.

Currently I'm playing Age of Wonders: Shadow Magic multiplayer and PBEM and I think there's nothing wrong with how they made it.

You can do simulaneous turns in MP games and you can set a timer on it, which makes the gameplay pretty streamlined.

PBEM is easy to do and great fun. I already played 'this way' back in the Commodore Amiga days via a 3,5" diskette. Conquest anyone?
Reply #12 from Stardock Forums Top
And now from the "crowd who never accepted this RTS/MMORPG/LAN network conspiracy" point of view:

I respect the fine tuning Galciv has gone through over the past months. There has been a lot of improvements and people know that I am not a whiner who asks for improbable things, but a bit of constructive criticism is required in regards to problems we have in single player Galciv now compared to the problems that would be inherited right now by a multiplayer Galciv.

The way I see it, why turn a darn fine single player game into a multiplayer game? Especially one that would rely on connectivity to a central server. Look at one of the biggest tech issues on the forum already: Metaverse submissions. If the Meta submission process is so confusing and even failing so often with individual players, what assurances does that give a player like myself who prowls the forum and looks at a thread with the word Metaverse in the title and sighs? Am I to think that a Stardock server can magically work on multiplayer turns, or even multiplayer score submissions, when there are so many recorded incidents of Metaverse problems on single player right now?

And then one would be opening the can of worms on difficulty levels again. Right now, it can be consistently proven in single player that scoring differences by difficulty level alone are pitifully underrepresented. Cakewalk is still scoring way too high in relation to even normal games. The scale is totally stacked toward people who play easier games. This problem would be magnified in multiplayer.

I think that player vs. player would be a better measure of skill and multiplayer submissions would therefore be better reflections of skill than the current single player system, but I do not think multiplayer Galciv's time has come until single player issues, especially with difficulty level scores and Metaverse submission, are worked out. Then there is also a little feature in place right now that would quickly spoil a multiplayer game: If I can find out in three seconds where the other player's home sector is at the start of the game from the influence map, isn't that a little bit of a problem?

I think talk of multiplayer Galciv is premature until we have a totally solid single player game that has a solid Metaverse. Then there will be enough proof that a strong foundation for server-linked multiplayer exists. I know that planning ahead is very important in gaming development and I do respect what has been done to fix the Galciv we know and love, but the very servers and scoring system that multiplayer would rely on are not ready for multiplayer in this player's opinion.
[Message Edited]
Reply #13 from Stardock Forums Top
I like the general idea Brad but like some of my colleages would suggest some changes.

Simultaneous movement: I believe that simultaneous movement is the best way to go if more than 2-3 players are involved. This forces all players to choose there actions and then simultaneously acts on them. A number of modifications would be required to the code, such as intercept, chase, defend options.

Game Flow: Each turn begins with the AI making all planned moves and changes. All ships will have predicted move icons showing where they are moving and what they are doing (attack, defend, move, repair). A list of build, tech, economy planned changes will also be displayed in the event window. The human plany now has X amount of time (from 1 min to infinite) to modify any of the AI's planned moves before clicking on the done button. They can also set global AI plans (focus on tech, military, influence), build queues, denense levels, etc. Once finished (or time runs out) ALL AI's move simultaneously.

I believe the above structure would solve most iussues with internet playing. A host of new code would be required but basically it would be the same game with humans able to control the AI. New players could jump in at any time (so long as an AI was uncontrolled). Old players could drop out. Games could be finished or paused and saved for a later completion date. The game could even play itself without human players if people just wanted to watch. Keeping the core game on the server is an excellent idea though.

Paul.



~SDC~
Reply #14 from Stardock Forums Top
The general idea of Brad is good.
Do you know the online game Ferion (www.ferion.com). ? This is a game in space, with planets, ships, research and everything as usual. This is a massive multiplayer online strategy game. The turns are turning each hour, this is a percistant game. You set your moves, start building or reasearching, you trade or sign treaties during your turn. Then, when the turn is resolving, the figths occur if ships are on the same coord.
Everything is on a central database and you acces it with a browser.

You could take the general idea for MultiGalCiv. The turns could append every hour (percistant game) or at a rate the player define (or when everyone has finish).

Have a look and tell me if I am wrong or not ?

Pask Hal
Reply #16 from Stardock Forums Top
And then one would be opening the can of worms on difficulty levels again. Right now, it can be consistently proven in single player that scoring differences by difficulty level alone are pitifully underrepresented
End of quote


That is an opinion, It cannot be proven. It can be proven that it is off relative to Frogboy's stated targets, but that's not what you typed.


The scale is totally stacked toward people who play easier games.

Is it? Then why this in the top 10?

Ray is 1, Maso 2 ebz beginner
Staffa 3, Maso 6 Wombat, beginner
Me 4. Painful. 9. Mauler Challenging
Jaws 5. Painful.
Heaton 7 Crippling
Treylis 8. Painful
vincible 10. Crippling

7 of the top 10 players play painful or above.
One plays challenging and two play beginner. I fail to see how that is stacked toward those that play easier games. Unless somone just doesn't wish to see a beginner crack the top 10 ever. I firmly believe that EBZ deserves to be where he is. Well ok maybe two spots lower :) Wombat too. They earned it.

I think, contrary to what some are saying, 7 out of the top 10 playing painful or above, and 8 being Challenging or above suggests that the system is NOT Stacked toward lower difficulties.





Reply #17 from Stardock Forums Top
Real time is definitely the way to do it. Chess is HUGE on the net these days...and after all its fundamentally very similar turn based setup too.

Each turn would need to be completed in X minutes. The participants should be given the option to decide how long that is when starting the game.

If somebody missed their turn, they would simply time out and the turn would be made anyway. No big deal really, its only a game after all. If not happy with your position, just start a new one. To avoid too many quits, you would always get some credit when playing to the end. Based on your performance of course, so just sitting there on it would not score much.

[Message Edited]
Reply #18 from Stardock Forums Top
Another option for timeout is to have the AI complete the turn for you. Have the player nominate which AI is their deputy during any absence or network lag. Maybe, and best of all, you could even program it a little bit on how it should behave in the typical situations.
[Message Edited]
Reply #19 from Stardock Forums Top
Another option for timeout is to have the AI complete the turn for you. Have the player nominate which AI is their deputy during any absence or network lag. Maybe, and best of all, you could even program it a little bit on how it should behave in the typical situations.
End of quote


One could choose (and presumably change) the personality (from several picks with sliders ) Agressiveness vs passiveness, builder vs expander, Techie vs build up military, etc of his chosen ai replacement should he be indisposed, offline, etc. And that would be really cool.

HEHE we're givin Frogboy alot of work to do...
Reply #20 from Stardock Forums Top
It would be great if it could be multiplayer, but GalCiv wasn't written to handle it. The effect of a turn based multiplayer would not translate well at all. When you move during your turn, it's an immediate action. Think of all the mean things you do to the AI that could happen to you. During someone else's turn you would watch your supposed friend attack with all of their ships and wipe out your fleet before you can do anything about it. Or the constant bringing up a diplomacy screen with them offering a trade. GalCiv won't work as a multiplayer game unless they change it and find a way to make it playable and enjoyable. I don't think Frogboy is asking us how we would want the game to play, but instead how the overall system (server hosting, easy player joining/leaving, ai filling in for players...) would work to make it feasible.
Reply #21 from Stardock Forums Top
Why not make it the first massive turb based stratagy game... lol, sounds funny... but sorta something like the Earth:2025 on swirve.games.com or whatevere it is. But have it all in GalCiv on a really big galaxy. That would be neat.... but an entirly diffrent game all together!
Reply #22 from Stardock Forums Top
It would be great if it could be multiplayer, but GalCiv wasn't written to handle it. The effect of a turn based multiplayer would not translate well at all. When you move during your turn, it's an immediate action. Think of all the mean things you do to the AI that could happen to you. During someone else's turn you would watch your supposed friend attack with all of their ships and wipe out your fleet before you can do anything about it
End of quote


Well I for one wouldn't let him get the entirety of his fleet that close to begin with. But They could just implement the civ rule. You must diplomatically break the alliance before you can attack and Breaking an alliance returns all units in the foreign territory home. (I forget which civ that is from)


I think we'd all know that the game itself would require some major changes. Hence, it would have to be a seperate game(albeit with galciv tendencies and stuff) all together.

But we can all dream can't we? Even though half the stuff I dream about in a game that doesn't exist yet wouldn't even be close to doable.



Reply #23 from Stardock Forums Top
Well I for one wouldn't let him get the entirety of his fleet that close to begin with. But They could just implement the civ rule. You must diplomatically break the alliance before you can attack and Breaking an alliance returns all units in the foreign territory home. (I forget which civ that is from)
End of quote


This is Civilzaton 2, which I actually still play.

~SDC~
Reply #24 from Stardock Forums Top
Each turn would need to be completed in X minutes. The participants should be given the option to decide how long that is when starting the game.
End of quote

Probably the best option for turn based, yes. Agree on starting time + time per turn, or fixed max time per turn and simultaneous turns.
Hosting the game on a machine (server being the player's machine or the metaverse) is a good idea, but...
Personnally, the only MP I am interested in is hotseat.

Reply #25 from Stardock Forums Top
Multiplayer games are sweet! I love beating the crap out of my friends... course... they never seemed to be too pleased. :notsure: But to be serious...

Frogboy,

Many have already tried to work out the specifics in this thread. I'm not going to lay out more details that would essentially duplicate their efforts. Instead, I’ll just skip to my conclusion. A fun multiplayer version of GalCiv is definitely doable, but only if you create a new game. I just don’t see the multiplayer option working out to be great unless it is designed and coded from the ground up as a multiplayer game. I would advise retaining only the plot and alien characters; indeed, I would demand it for continuity! Game mechanics, the Metaverse, and even the whole ‘philosophical’ approach would have to be completely redone. If Stardock produces a multiplayer GalCiv based on such efforts, I wouldn’t hesitate to buy it... worship it... make animal sacrifices at its altar... etc. :)