The critical mass problem

Draginol · February 8, 2003, 16:55

One of the problems in most strategy games is the later stages where you KNOW you've won but you have to mop up.

We try to avoid that sort of thing in GalCiv through a variety of mechanisms. But no mechanism is as straight forward as common sense.

In most strategy games, the AI opponents tend to gang up on whoever is winning. Hence, if you're winning the game, even your closest friends will slowly move away from you.

Nothing causes critical mass problems more than that. And it is, IMO, a very very BAD game mechanic. In the real world, it works the opposite. The more powerful you are, the more other governments will tend to want to get along with you. Sure, they may snipe at you or whatever but they're not likely to go to war with you.

GalCiv works along those lines. If you are winning the game, your friends remain your friends. In fact, they may want to move beyond friends and become your allies.

As a result, mop up usually is a minor thing in the game because AI civilizations who are hopelessly outgunned will beg for mercy and want to eventually become your ally even. Failing that, they'll surrender outright.

But as we were playing this week, it really became apparent just how much of a problem it is when AI designers decide to have relations based on inverse power -- i.e. everyone wanting to gang up on who is most powerful. If game designers would just not do this, much of the issues with critical mass would go away.

MrBaggins · February 8, 2003, 16:59

Read this

moomin · February 8, 2003, 18:03

I dunno. While there's a problem with critical mass, Roeynold's solution - punish the leader, help the laggard - isn't really what I'd like to see in a 4x game, which is all about becoming the leader, after all. Perhaps MP game balance requires it, but a game like GalCiv, being exclusively SP, can do better: acknowledge a victory and get on with it.

MrBaggins · February 8, 2003, 18:11

He deals with these concepts, in the last two paragraphs of 'Steady Progess'

MrBaggins · February 8, 2003, 18:16

But... SP games need challenge.

The critical component of the game remaining fun, is the maintainance of that challenge...

Diminishing the elastic rewards/penalties that keep things close longer ultimately hurt competitiveness and ultimately diminish the fun gained.

Games differ from reality in that their purpose is the enjoyment of those that hold the strings... not slavish devotion to 'reality'.

Not that there's anything wrong with realism... just that it should take a back seat to fun... when it comes to games.

MrBaggins

Asmodean · February 8, 2003, 19:06

As I see it, Reynolds and Wardell want essentially the same thing: To make a fun game. They just approach the thing differently. Reynolds' method would probably tend to prolong the game, wheread Wardell's method would tend to finish the game early. Nothing wrong with either method.

Asmodean

Draginol · February 8, 2003, 19:12

Well put.

I don't agree with the "punish the leader" system taken to the nTH degree.

If all the other players combined are much more powerful than the leader, then yes, they should see about thwarting the leader -- to an extent. If someone is allied with you or already has very close relations with you then they should remain loyal.

But if all the players combined cannot even come close to the human player, then it's over.

The AI *should* know when it's hopeless.

Example from today's game:

http://www.joeuser.com/images/beta5/3.jpg

Draginol · February 8, 2003, 19:18

Incidentally, I wonder what the statistics are on someone FINISHING a Civ game via conquest on a very large map.

In GalCiv, the gigantic galaxy size is 240 moves by 240 moves. That's a HUGE map to deal with.

I can't imagine how it would be "fun" to control 90% of the map and realize that you have to mop up for another 2 or 3 hours when victory is completely inevitable.

Now, if you only control say 40% of the map and each of the other 5 players only control 12% then that's a different story. Each of them could work to try to thwart you if they're not already allied to you or very close friends to you.

The scenario that I personally have a problem with is where you control 40% of the map and are allied to 3 of the players who control an additional 30% when combined (70%) and then for no apparent reason they a) break their alliances with you and b) eventually go to war. No way. That's ridiculous. It completely undermines those who want to be the master diplomat.

If I'm at 40% of the map and the other guys are all neutral to me, then they can and will conspire against you. Similarly they will conspire WITH you against someone else if they're in the same situation.

But when someone controls 90% of the map it's over. It's time to reward the player and expidate the end of the game so that they can move on to other challenges.

MrBaggins · February 8, 2003, 19:22

True... but *EVERY* effort should be made to ensure that the situation doesn't get to that point... THAT is Brian Reynolds point...

He actually pointed out that the game should end instantly... at the moment that victory was assured... or defeat certain.

Delay... and continuation of a viable challenge, is desirable, if the game experience is enhanced by late game technologies or situations.

If the late game technologies, are just same-ish... then I guess it doesn't really matter.

Since modern techs make for such different war possibilities... in an 'Earth-like' technology situation, you should want to stretch the game out.

MrBaggins

Draginol · February 8, 2003, 19:28

Are you suggesting that Civilization 1, 2, or 3 (particularly 2) don't have serious critical mass issues?

MrBaggins · February 8, 2003, 19:33

yes... they obviously do.

ICS reigns supreme, and at the other end... and at the other end, large 'cities' are positive-feedback loops.

I'm fixing this... in a game engine that allows the game defects to be fixed.

Harry Seldon · February 8, 2003, 19:39

Quote:

He actually pointed out that the game should end instantly... at the moment that victory was assured... or defeat certain.

I don't believe I'd want my game to end just because the AI feels I can't win. I'm not through until I see the transports screaming through the atmosphere of my last system.

I'm glad Stardock is addressing the critical mass issue and feel they've taken a good approach to it. I've never played Civ, but in MOO it got very tiresome MMing hundreds of planets trying to find that last colony ship the Darlocks sent to BFE right before I slagged their last planet.

MrBaggins · February 8, 2003, 19:42

Quote:

I don't believe I'd want my game to end just because the AI feels I can't win. I'm not through until I see the transports screaming through the atmosphere of my last system.

That should be achieved by them massively ganging up on you, and giving you no quarter... as you fight to your last man... *smiles*

LDiCesare · February 9, 2003, 08:35

The big difference between the approaches lies in the effects on gameplay:
By thwarting every player's moves when they become stronger and promoting ganging up against a leader, you make it very hard for someone to ally themselves with someone else, and thus promote more war-like behaviour. That, IMO, limits the number of options. In contrast to Reynold's approach, GalCiv promotes various ways to win. I doubt they could do it with a "punishing leadership" route. In particular becasue you may want to lead in one area and leave the others undevelopped, but if you are penalized for being leader in a single area (like tech in civ), then what's the point in specializing?
I feel punishing the best means punishing good gameplay. Games should provide challenges by providing choices. Additional choices mean you can't lead in all areas at once. If you do and have contact with all opponents, then you have won and the game should be over soon. ICS is a problem here in fact because it allows to increase every area of the game (population, trade and production all at once for little cost). Big cities are less of an asset if confronted to a good opponent, because you would face pillaging which effectively destroys the city effectiveness.

Draginol · February 9, 2003, 15:06

harry: The human player can play to the bitter end. But the computer players will generally resign if they think it's hopeless. Or more to the point, they surrender to a player.

They may surrender TO you or if there's someone they think has a chance to defeat the leader they'll surrender to the toughest opposition to try to strengthen them.

But what you don't usually have to deal with is the scenario where you control 300 planets and your opponents control a dozen across the galaxy.

When the player has won, the game should acknowledge that. Not just because it moves things forward more quickly but because it's fun to see computer players say things like "You've crushed our civilization, we beg for mercy and surrender to you and your greatness!"

We want groveling!

Or as the Torians so meekly put it when you talk to them in situations like the above:

"I don't feel I'm even worthy to speak to such great and powerful beings. But I will try my best. How may we humbly serve you?"

Asmodean · February 9, 2003, 15:28

I remember several Alpha Centauri games where winning was a foregone conclusion, but mopping up still took a lot of time. I like the fact that this issue is being adressed in GalCiv. One thing I would like to see in GalCiv, though, is a feature like "Resign". Civ3 has it, Alpha Centauri had it, and I think it would be great here too. Sometimes you just find yourself wanting to see the endgame stats without having to wait for the endgame.

Asmodean

notyoueither · February 9, 2003, 17:23

Two different approaches. It must be possible to make a good game by pursuing either. You just have to skin the cat differently.

And yes, I think I will appreciate a game where I do not have to conquer every last outpost in order to win.

Although, it might be a good idea if some of the races would be more determined to hang on and slug it out than others. Is this the case?

Harry Seldon · February 9, 2003, 20:21

Quote:

harry: The human player can play to the bitter end. But the computer players will generally resign if they think it's hopeless. Or more to the point, they surrender to a player.

I did get that from reading about the game. From Mr. Baggins post, I thought he was supporting the idea that a game should stop when winning is hopeless and wanted to state for the record that I'd like the choice to play on. That's one of the many reasons I'm drawn to GalCiv. It would be cool if you had the option to resign or surrender to your opponent though. Last stands are all well and good; sometimes I'd like to avoid the utter humiliating defeat.

MrBaggins · February 9, 2003, 21:00

I honestly think that this whole 'surrender' thing is missing the whole point.

Players in TBS' know when they've won, and don't need a 'we give up' message to be informed of the fact.

The AI is there to give you a challenge as long as you demand that of it... I.E. as long as you continue to play.

MrBaggins

Harry Seldon · February 9, 2003, 21:07

But the AI can challenge you anew when you start a new game. If I know I've won, many times I'll just bail out of the game and start again without finishing it. Who wants to spend three hours cleaning up a forgone conclusion? I'll admit, sometimes in MOO I'd enjoy conquering every system but after the first few times I would have welcomed a Mrrshan surrender instead of killing them to the last man(cat?).

MrBaggins · February 9, 2003, 21:20

So... ultimately what is the point of this?

You get a message stating... "you've won" rather than your knowing you've won, and starting over yourself?

Big deal...

How is this an innovation?

LDiCesare · February 9, 2003, 21:51

It is a big deal, because you know that you have won, not just believe it. Plus, you get whatever reward you can get for ending the game: end video, stats, and the ability to set up a record or high score if you are interested in that.
If I quit a game because I know it is won, I can't help feeling a sense of "uncompleteness" or something. Without it, I'd have a better playing experience.

MrBaggins · February 9, 2003, 21:55

Its never really over 'til its over though:

The French surrendered in WW2... should they have stayed surrendered, in your view?

Harry Seldon · February 9, 2003, 23:11

The French surrendered because they had no choice. They were beaten and had to do what they could to save their people. Hence, the AI abdication in GalCiv. When there is no hope for victory, an intelligent species will ensure it's future, just like the French and just like in the game. After all, if there's no one left there's no chance of a return to power.

ravagon · February 10, 2003, 01:29

Quote:

Originally posted by Harry Seldon
The French surrendered because they had no choice. They were beaten and had to do what they could to save their people. Hence, the AI abdication in GalCiv.

True but (a) It wasn't an unconditional surrender (the Vichy puppet continued for a time) and (b) The French didn't know (or didn't want to know) what they were getting in to - ie: How France and the rest of Europe would fare under Nazi rule.
GalCiv seems to include the "reaction-adjustment-through-experience" factor - ie: Based on the ethics of your actions through the game up to the point.
It would be nice if the surrender option factored this into the equation - ie: Ranging from essentially unconditional surrender (to an honourable opponent) to a no retreat-no surrender option for a particularly heinous empire guilty of exterminating/enslaving previous conquests.

Jamski · February 10, 2003, 04:36

Well, as long as its not like in SMAC where my submissive pact brothers would go to war with each other

or try to steal my tech

Then I'll be happy.

-Jam

Asmodean · February 10, 2003, 11:35

Quote:

Originally posted by MrBaggins
Its never really over 'til its over though:

The French surrendered in WW2... should they have stayed surrendered, in your view?

Not a good example. In gaming, when you control 80% of the cities/planets/marshmallows/whatever, and you largest competitor control only 9%, the game is won. In WW2, there was still USA as the unknown factor, when France surrendered. Then, when given the chance by the U.S., France stepped back in to the fray. In Galciv USA will not suddenly materialize and help the AI. Trust me...It won't happen.

Asmodean

Draginol · February 10, 2003, 14:15

I suspect I am not the only one who has "retired" (and greeted with a "you have been defeated" message) in a strategy game despite knowing that we would eventually win the game.

If I have 300 planets/cities/whatever and my opponents have 12, it's over. The only way it ain't over is if some incredibly cheesy random event came up to sabotage it and that would definitely not be an improvement IMO.

But often times, I don't want to spend the extra hour or two to mop up those last 12 planets. At a certain point, the AI should: a) Try to suck up to you to become your friend or b) surrender to you.

Now admittedly it's not always going to work out in those two cases. If, for instance, you are "evil" and the alliance of races taht make up that last 10% are "good" then they will fight to the bitter end. That's the downside of playing as evil, the good guys will go on a Wilsonian crusade despite the odds.

But in many if not most cases the AI will retire.

Which brings up an irony, those who love multiplayer surely remember playing games against people who play "hide the last unit in the corner to prolong the game". Having an ego-less AI does have its advantages.

Harry Seldon · February 10, 2003, 14:37

Quote:

Now admittedly it's not always going to work out in those two cases. If, for instance, you are "evil" and the alliance of races taht make up that last 10% are "good" then they will fight to the bitter end. That's the downside of playing as evil, the good guys will go on a Wilsonian crusade despite the odds.

That's as it should be. An AI that truly emulates intelligence would choose to follow the path of salvation, either by surrendering to a power to preserve their people or like in your example fighting a way of life that will surely lead to the destruction of their people regardless.

Man, I wish I was a beta tester!

Sandman · February 23, 2003, 16:00

Having the other races give up and join you is a much better solution than imposing artificial barriers on successful playing. I think that such 'elastic' solutions are simply unfun, chaining the player for no good reason. Part of the fun in civ is 'the thrill of the chase' in catching up with rivals, or struggling to keep your lead.
I think that the player will be more creative (and thus have a more enjoyable playing experience) when the pressure is on, so to speak.

Having said that, it's absolutely imperative that it be difficult to make competitors give up and join you. I can see a snowball effect, where you and an single ally get to a certain critical mass of power, from which all other races agree to join you.