'Puzzle' v. 'Game'

[F_Smith]

Interesting topic came up in the govt thread:

Q.

[*]Should a player's actions always succeed?[*]All other things being equal, in a combat, should the same unit win every time? [*]Should a Panther Tank *always* defeat a Sherman tank one-on-one? [*]Should a Musket always do X points of damage? [*]Or should there be a random fluctuation in the system, allowing the improbable to happen?[/list]

I thought I should explain myself fully to everyone. My idea of how to make a game seems to be somewhat 'unique', here. I seem to be a lone voice in the wilderness.

I believe that a 'sim' is actually more of a 'puzzle', than a game. Caesar 3, Pharoah, Railroad Tycoon 2 -- playing those is a matter of figuring out the 'if-this-then-that' puzzle. Once you've mastered that, 'figured out' the puzzle, the game becomes repetitive. If you play a scenario in exactly the same way, it will turn out exactly the same. If two players do exactly the same things, their civs turn out the same.

This is where the emphasis on 'build order' comes from. 'Build Order' is the dominant game concept these days. From so-called RTS's to 'Civ' clones, 'build order' passes for strategy.

It has become clear that many here want a 'sim' type of game for Clash. I push my 'optional' rules systems because I want to avoid Clash becoming a sim puzzle that people tire of after they've learned how the equations react.

I hope that explains, somewhat, why I disagree with some of these models. Why I push for flexible 'object' models over equations. So just be patient with me. I've been making games on paper and computer to play with friends for 20 years, as have many of ya'll, so I'm pretty experienced at actually building games. This is not an 'arbitrary' thought I just came up with.

[Richard Bruns]

How many people here have studied chaos theory? I have, and I think that my studies have given me insight into the behavior of game systems.

The basic idea of chaos theory is that a simple deterministic model can become very unpredictable when iterated many times.

When you iterate a function, you plug the result back into the input side of the equation. If you keep doing this, you can quickly get wild results. In fact, an infinitesimal difference in the input can result in an infinite difference in the eventual result.

For example, think about the equation x=(x^2)-.5

This is a very simple equation, but its behavior gets very complex when you stsrt experimenting with its iteration. To prove my point, I will bore you by describing a very small part of the complexity of this equation:

Let's start by iterating some values. For the first example, I'll choose 1.5 as the initial x value and iterate, rounding to four decimal places:

x0=1.5
x1=(1.5^2)-.5=1.75
x2=(1.75^2)-.5=2.5625
x3=(2.5625^2)-.5=6.0664
x4=36.3013
x5=1317.2833
The eventual value of x will be infinity.

Now look at another value:

x0=1.25
x1=1.0625
x2=.6289
x3=-.1045
x4=-.4891
x5=-.2607
x6=-.4320
The eventual value of x will be an irrational number whose first ten digits are -.3660254038.

As you can see, some initial values of x will result in infinity. For other values of x, the eventual value is a fixed number. For two initial values (the fixed points), the output is the same as the input. And for a lot of other points, the system settles into a cycle.

In this simple example, it is possible to find the fixed points mathematically. I already described one of them, and the other is an irrational number whose first ten digits are 1.366025404. No value will go to that value, but if it is input, the equation will output that same value. This value is a border between points that go to infinity and those that go to the fixed point.

I could keep describing different things about this equation for pages, but I think I have made my point. The simple equation can be complicated and unpredictable. Unless you know the math and do a lot of work, you cannot predict which values will go to infinity and which ones will go to the fixed point.

Now consider the equations in Clash. There are hundreds of them and they are much more complex than the equation detailed above. They will interact with each other in complex ways every time the system is iterated (when the "End Turn" button is pressed).

I think that trying to predict the events in Clash will be like trying to predict the weather. You might be able to predict a few days, or turns, in advance, and you might see big storms coming from a longer way away, but beyond a certain time period all predictions will be useless. You have probably heard the example about a butterfly somewhere being able to mess up weather predictions all across the world. It has been proven that the tiny effects of the butterfly flapping its wings can change major storm systems on the other end of the world.

That comes from chaos theory. And it is my opinion that there will be thousands of these "butterflies" in Clash.

Unlike other games, there will be a lot of things going on that the player does not see or cannot influence. The ecology model will alter the landscape. The social model will change the way the people think. The tech system will produce surprising and unexpected results because it is much more complex than the simple x RP's=new advance systems of the past.

I'll concentrate on the tech system because I think that it best illustrates the difference between Clash and previous games. In Civ 2, techs were bought like any other commodity. You paid a certain number of RP's and you got an advance that was fully operational and did not change for thousands of years. Nothing had any impact on them once you got them. This was simple. This was boring. This allowed players to create build orders for their techs so they could climb the tech tree the same way every time.

Our tech system is different. All of the techs are changing at the same time, and lot of the tech change is not controlled by the player. Social forces, military actions, economic activities, climate patterns, and natural disasters will all have subtle influences on the technology levels. As the system is iterated, these "butterflies" will be able to have a big impact on the technology the player has access to. The player will never be able to predict exactly what will happen. The player will have to react to the unpredictable effects of these "butterflies" that iterated themselves based on simple equations.

In other words, no player will ever be able to create a build order for our tech system. They will not be able to control it like they could control Civ 2's tech growth. While you can usually predict what techs a good civ 2 player will have at a certain time, I can guarantee that you will not be able to predict where the techs will be at any given time. Chaos theory shows that a difference of 1% in a single social attribute could result in big tech changes after the game mechanics are iterated 300 times.

And all of this happens even though there is almost no randomness in the tech system. Simple equations will produce complex and unpredictable results when iterated hundreds of times.

My point is that we do not need to add extra randomness and chaos to the system. It might seem like the models are simple and predictable equations, but they are not. The interconnections between our models will cause "butterfly effects" that will produce more chaos and randomness then any other game in existence.

[F_Smith]

Richard:

Yes, I have studied 'chaos' theory, as probability is now fashionably called. OO programing, as you might have notices, works hand in glove with 'chaos' theories.

And the funny thing is, the basic mantra of 'chaos' theory is that the outcome of even the most complex systems *is* predictable, as far as producing patterns. You can find order in chaos. That's where the name comes from, in fact.

That's what's biting us here.

Look, ya''l, we have experience to work from. Clash is not going to be as complex as Caesar 3. Caesar 3 becomes repetitive. Sim City is more complex. It gets repetitive.

Please, tell me -- have ya'll played a 'sim' city type-game? If so, did you find it repetitive, after you 'figured it out'?

[F_Smith]

Another excellent example:

Have ya'll ever played Settlers 3?

There may never *be* another game as complex as that one. And yet it's *very* build-order oriented. It becomes very predictable, once you've played a few hours of it. It's still a fun puzzle, but not really a game.

My idea of a game involves 'conflict'.

Two sides with different goals. They can either work together, or against each other.

But don't get me wrong, if that's the kind of game you're after, we should certainly leave the 'default' Clash game a sim-Civ thing. Then, as options, we can add versions of each model that are more 'game' than puzzle.

[Richard Bruns]

Yes, I have played SimCity. Clash will be a lot more complex than SimCity ever thought about being. Consider:

SimCity had no technology change.
There was no military conflict.
For practical purposes, there were no other cities or governments.
The landscape did not change, and food production was not considered.
There were no religions, no social groups, and no general patterns of thought in the population.
There were no characters or personalities.
It only had a crude representation of the economy.

There is a big difference between detail and complexity. SimCity was a very detailed representation of a relatively simple thing. The player did a lot of micromanaging of little things, but the scope of the model was extremely limited and there was nothing happening behind the scenes that could have a big impact on things. With the exception of the disasters, everything in SimCity happened as a direct result of player action. There were no "butterflies".

There was nothing like social attributes that evolve over time and affect the technology tree, which affects economic activity, which affects the country's ability to wage war. There was nothing like the gradual climate change that affects farm yields and forces migrations. There was nothing like the wars and conflicts that have long term impacts on the economy and society.

The other games are also like SimCity. They are very detailed models of simple things. There is no scope and no sense of long term events. They are puzzles because the player is responsible for everything. In Clash, we have thousands of things happening that the player is not controlling.

Yes, there is some order in unchanging equation systems. But if things are constantly being altered, you have a situation where you will never know exactly what will happen. If the -.5 in my equation above changed periodically, you would never be able to figure out exactly which values would go to infinity.

The weather is like that. Sure, they may know that hurricane season is a certain time, but no one will be able to predict a month in advance where a hurricane will hit.

If all of the models remain as complex as they currently are and we do a good job of connecting them together, Clash will be like the weather. The details will be completely unpredictable over the long run.

[F_Smith]

Richard:

Sim City 3000 is actually a *very* complex interaction of several 'systems', including other 'cities', terrain, technology advancement, an extremely complex econ and trade system, and on and on. It's far more complex than *anything* in Clash, and I don't mean just detailed . . . altho it is that, too. Believe me, it's really *too* complex. It's not all that much fun (to me), after the first few weeks, when the 'new game' sparkle has worn off.

And then take Caesar 3. Even more complex than Sim City. Includes other nations, terrain, combat, religions, social classes, even individuals and characters. Still predictable.

Then take 'Settlers 3'. Even more complex. The econ supply chain is amazing! You have to get a woodcutter to build a plow for the farmer to grow crops, and the like. The complexity is astounding. But still predictable.

Railroad Tycoon 2 has all the same things (except combat) -- tech, characters, terrain, other countries, other railroad companies run by other characters, a 'stock market', a 'bond market', banks, etc.

Adding complexity didn't change the level of predictability in those games. Why would it be any different in Clash?

Perhaps you're not considering the effects of replay. You don't do the math in your head, you 'test', thru gameplay. The way you 'predict' what will happen is by playing a scenario, doing x, watching the results, then replaying and doing y.

After you've played with the game for a few weeks, you will know exactly what to expect from the govt policies. Mark has what, 200 saves of Civilization on his PC? After that kind of play, Clash's default govt system will be *very* predictable.

By thinking no one will 'master' the system, you're not taking into account how smart gamers *are* (and how much time some of us have on our hands). There will be some 14 year old in Lithuania that will run those tests and figure it all out, then build a web site showing charts and graphs of exactly what to expect and when, complete with build orders and scenario walk thrus. Just like all the other, highly complex 'Sim' Civ games.

And it may be my personal opinion only, but -- if someone can write up a 'build order' for a player to follow in a scenario ("First, do this. Then do that. Then do this."), then the game is too predictable.

[Richard Bruns]

The difference is that in Clash, the player does not control everything. In the other games, you told everybody exactly what to do and they only did what you told them to. They were not independent entities interacting with things that are not controlled by the player. Everything was just a little puppet that was manipulated by the player in a big, planned puppet show.

In the other games, the player controlled almost everything, which meant that if the player did the same thing, the game did the same thing.

In Clash, things are independent. The market makes its own decisions. The population have a direct influence over tech growth. Little things will have a chance to grow and change without the player becoming aware of them, and then they will suddenly pop up and demand attention. Even if a player does exactly the same thing, the independent things in the game will do different things and create a different game.

The other games put the butterflies in jars and attach strings to their wings. The player can manipulate them and the air currents never get out. But as far as I can tell, we have them all flying loose. They are not controlled by the player, and the impact of their actions can spread unhindered.

Basically, we have a dynamic world and the other games had a theater. They relied on the player's script while our models will run without a lot of player control.

Of course, if we do things wrong we will have what the other games do. But if there are enough interconnections between our models, we will have all of the chaos we will need.

[F_Smith]

Richard:

You're mistaken.

The other games are *exactly* what you're trying to build.

Caesar 3 is one of the neatest, in that regard. Everything is independent of the player. Your citizens walk around living their life without you. All the player can control is building buildings and setting government policies. The people in your city wander your roads on their own, looking for the things they want (jobs, access to religion, consumer goods like food, etc).

Railroad Tycoon 2 is even neater -- there's an 'economy' that operates without you, creating booms and depressions, altering the course of everything.

I just don't understand the idea that complexity equals uncertainty. If the system has no uncertainty, then it will be predictable to someone that has experience with it.

[Richard Bruns]

I know about Caesar 3. Those sims don't really do anything and they have no power. They basically have a wish list. They will either consume products or leave. They don't build things themselves, they don't start revolutions, they don't belong to social classes trying to get power in the government, they don't influence tech research with their opinions, they don't really act human. They are like ants in an ant farm. The player is not necessarily dictating their moves, but they are being controlled and coralled just as neatly as a herd of cows.

I never said that there was no uncertainty in Clash. The ecology model is a big jumble of random events and processes just waiting to mess up things. The disaster model is about the same thing. I thought there were a lot of random fluctuations in the social and riots models as well. The population model will have randomness, in addition to being affected by the randomness of the social model. The economy and technology models are intertwined with all of these, so they will both be affected by all of these random events. The game will be chaotic because the effects of all of these random events will spread throughout the entire system.

I agree with your comments about player intelligence. I have been saying for a long time that we should eliminate micromanagement opportunities so the players don't have a chance to manipulate things. For example, I wanted the foul play in politics to happen randomly, without direct player control. I wanted province boundaries to be fixed so players cannot fine tune them. The other games let players control too many things, which means that the game is controlled by the player.

My ideal of Clash is a game that forces players to react to changing circumstances and does not let them impose their rule on the world. My models mostly run on autopilot. I made sure that the tech advancement operates mostly without the player, just as it did in real life. (Players can only add a few RP's to supplement all of the things that are done automatically.) The ecology model is almost entirely uncontrolled by the player and the population model also runs automatically, based on info from the social model that I assumed would do random things.

If the player must react rather than act, fine control and build orders will be impossible. I think that our models are independent enough and random enough to be chaotic and force the player to react.

But while I work for long term chaos, I don't like short term unpredictability that forces players to try to micromanage and fine tune their policies. The player should feel steady and in control over the short run, while long term chaotic forces are lurking beneath the surface waiting to strike.

[axi]

I intended this to be part of the other post, but this thread is more appropriate:

Determinism vs randomness: I am the prowd owner of SimCity 3000, which I don't play, for one reason because I couldn't find the right solution to the "Puzzle" (so my cities always either stagnated or plunged me into debt), but chiefly because I ran out of disc space and I had to uninstall it (along with CtP) to make space for Baldur's Gate (which was definitely worth the space). I have greatly enjoyed the Caesar 3 demo and have also played the Settlers 3 demo. The "winner formula" is always present in strategy games (though less in RTS warfare games, where tempo is the chief issue) and I can cite even more extreme examples, such as the Heroes of Might and Magic series (which is tremendous fun nevertheless) and the Imperialism series. In civ2, there are more than one winner formulae, but they are also limited. What is bad about civ2 is that, in contrast with other games, winner formulae tend to be based in extremism instead of balance: it's all caravans, or all spies, or all howies, or all settlers: if you play all of them, you will probably lose. We need to take care of that too; it's more interesting to try to achieve some balance. I don't think that the existance of a winner formula has anything to do with a chaotic or deterministic system. Even in a chaotic system, there are solutions which are safe enough to be repeatedly implemented; whether they are always optimum is another issue.

Randomness should exist where you'd expect it. A phalanx shouldn't have the slightest chance to bring down a stealth fighter, but a WWII AA gun should. But F_Smith is right about saying that some games have butterflies into jars. The most recent discovery made at the civ2 strategy forum is (coincidently appropriate with our discussion) that an instant revolution (with 0 turns into anarchy), can and definitely will always occur in a certain sequence of turns that they have found out and printed in a table to be used as reference. If you think about it, the thing is ridiculous: would it be so difficult for Sid Meier to make the # of turns in anarchy random? The same applies to science beakers and many other things in the game.

But Clash stands very high above this. The riots model f.e. which is in NO WAY deterministic. Determinism in it ends in the computation of events probabilities. If these events will occur (and how serious they will be), it is left to chance. It is expected to be a great randomiser of the game. So will the sophisticated AI, wherever it enters (event into politics, if we wish). Add this to complexity and you'll have a good game. As a "good game" I regard a game that is no more and no less chaotic than the real world. Order exists inside the chaos and the player reads it's patterns, that's how he can play in the first place.

[Mark_Everson]

Oops! Somehow I got the half-baked notion that the riots model was deterministic. Of course it isn't, its the Govt model that is deterministic. Thanks for setting me straight Axi

F_Smith:

I simply do not understand why you won't accept the solution that if things in Any model are too predictable (so that its not Fun) we can simply add some randomness to the model.

All:

As a matter of fact it might be an interesting option to give the player to let Them decide on determinism level. All models could then have random elements baked in (most do already). And the player could be allowed to select the amount of random component in each on a 0-100 slider or something. Generally you would want more randomness in models you are interested in as a player. Go ahead, flame me...

[Mark_Everson]

Well, I'd already responded on this topic in the govt. thread, but here's a bit of it:

If the player can always predict what can happen then the model is a failure IMO. (At least if its an important part of the game.) But I think the deterministic riots model is so complicated it may defy prediction anyway! I don't think determinism is such a sin in a complicated model, because the chance of Ever being under the exact same conditions is vanishingly small. If it is too predictable in a way the player can exploit, then I would vote strongly for putting randomness in it. But that is easily done if we decide its needed.

Equations vs Objects IMO has Nothing to do with it... Equations can generally be made to have a random component with very little effort.

[F_Smith]

I'm sorry, I didn't see this post, I replied in the other thread.

Should we continue here, or there?

P.S. -- equations v. object model has everything to do with it, I think. Because it's a mindset. A 'sim' game is just a big equation. A modern 'computer game' implies a much more flexible, object-oriented system.

[F_Smith]

Richard:

Caesar 3 is *very* complex. It's a good puzzle. It's a world that runs on autopilot, in which the player as ruler is constantly reacting to changing circumstances and has little or no direct control. It is very much like what you seem to be describing. You ought to play it. The people aren't stupid, they act in many ways very *real*. It's a cool, good sim. I hear 'Pharoah' is even better, too.

As I said, I believe a great game has turns based upon conflict. And that conflict *requires* short term, turn-by-turn unpredictability. So I built this model for running the govt, based upon those criteria. I, personally, like it very much. If you, or the others don't, I respect that. Ya'll should probably not use it.

I like to have control in the game. I wanted to write a politics game in which you had more 'realistic' control (within reason) of the government.

* * *

Axi:

I hope I didn't give the wrong impression -- I do not feel Clash has no randomness. My comments are meant to be about the govt policy selection process only, in this case. In rambling and raving I may have made it sound like I was complaining about more than I really was. Other parts are fine, I'm sure, and I know the riots model delivers probabilities.

I'm only talking about actually running the govt, on a turn-by-turn basis.

* * *

Mark:

That is the currently accepted solution.

I have not demanded any changes to the default game at all. I'm perfectly content to leave the default game as ya'll designed it.

This is all just about my 'optional' system, which seems to have invited a firestorm of excellent constructive criticism.

This is just about something that I found fun. A system seemed to build itself when I did the analysis, so I wrote it up alongside the 'default' one. It seems to work extremely well, and I find it very fun. It probably is only a personal preference, no one here seems to find a politics game fun at all. I am probably the only one that would enjoy a game about policy negotiations. But that's suppose to be the beauty of Clash, right?

I want to play a ruler, and set the military on autopilot, I can do it. You want to play a warlord, and set the govt on autopilot, you can do that.

I built a govt negotiation game because I want to play a govt negotiation game. Same reason ya'll are working on Clash, I suspect.

[Lord God Jinnai]

Well we don't haveto go and add randomness yet until we figure out if its needed. I understand the need for randomness, but i also understand the need to have some idea what's going to happen.

Some models really can't work with randomness, like the Tech model, while others, like the ecology and disasters model can be a lot more random, the latter almost completely, cept with the global warming/cooling/whatnot effect