'Puzzle' v. 'Game'

[F_Smith]

Lordy:

Besides, 'randomness' isn't really the goal.

The way to make a great game is 'conflict'. That's how you create unpredictability -- the outcome of a turn is based on the choices that two different sides make.

Adding 'random' events to a game usually annoys me. I hate it when a 'random' event sets me back in these types of games.

[Simpson II]

I just stopped by, and wanted to make a few comments

For a start, I'd like to point out to F_Smith that Chaos Theory is not probability, nor in any way related to probability theory. Furthermore, it does not say that "...the outcome of even the most complex systems *is* predictable." Chaos theory deals only with systems where all of the mechanisms are known and completely deterministic - that is, theoretically predictable. It simply ignores non-deterministic systems.
It does show that minute differences in starting conditions can change the eventual condition of a system completely out of recognition, and therefore that given non-perfect data about the condition of such a system, predictions of it's future state will grow increasingly inaccurate as time goes on.
It also showed that there were systems where, for a certain range of starting states, the system will not;
1) Converge or diverge, that is it will neither approach a given state continually nor go towards infinity (where infinity is a relevant concept.) Richard's sequence is an example of both these types of behaviour, for different values.
2) Repeat it's current state.

Ironically, the pseudo random number generators that would provide the 'random' element you are looking for, F_Smith, work on precisely these principles, so you are stuck with Chaos Theory one way or another.

Sorry if that bored you, but I wanted to set the record straight. There's too much bad pseudo-scientific literature around these days.

Second, Richard said:
"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.... 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."

Errrr... the game being controled by the player, eh? Terrible. Ummm... but isn't that why we play Civ and Smac? The idea of a player reacting to a series of random problems might be pleasing to a designer, but it's not going to be much fun to play, at least not for the average Civ fan. For me, the idea is to build an Empire which will resist the external forces arrayed against it. Look at how agravating things like the AI entity and empire-wide revolts due to running out of gold are in CtP! If the player doesn't control their own empire, then the game has absolutely no relationship with Civ, and I humbly suggest coming up with a new title.

axi, you said
"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."

If the formula is balanced between all of these, then I think you will end up with one optimal formula. Extremism must be an option, otherwise Adam Smith's rants about the Law of Diminishing Returns become true, and you end up with a rather pedestrian, optimal strategy.
Inevitably, it's easier to play an extreme game because it involves less decisions, and usually far fewer really critical decisions. In my experience you can achieve a balance between the styles in Civ and especially in Smac, but you have to be good. Over on the Smac boards it's known as playing hybrid style. Overall it is potentially more powerful than other styles, because they usually encounter a percentage of 'unwinnable' games (in MP.) Only a hybrid can expect to succeed - or loose - from any position.

On the whole randomness issue, in CivII/Smac the chaos comes from the interaction between players. Yes you control your own Civ/faction, but you don't control everyone else. And, because of the fog-of-war, you don't really know what they're doing at any given time, either. IMHO there should be a little randomness in the game behaviour, but largely it should be in the AI, the diplomacy, the combat and the terrain. I don't usually harken back to bygone achievements, but what you guys seem to be suggesting sounds very wrong to me. OK, the Civ/Smac AIs are inane, but I'm sure you can do better.

[F_Smith]

Richard:

But again, the system isn't 'random', it's 'outcome is produced by conflict'.

That introduces uncertainty into the game, I feel, without randomness.

The Politics game's outcomes are not 'random'. You will have control over that tax increase. You can also choose randomness, take a chance, offer a smaller bribe, that sort of thing. But even then, you still have control over that randomness, and can decide exactly how big a chance to take.

Similarly, the 'riots' model's outcomes are not purely random. You have control over the probabilities.

I was talking about purely 'random' events like the weather, earthquakes, etc.

And yes, that's the same Caesar 3 we're talking about. It's a Roman Sim City.

But again, Sim City isn't quite the 'fluff' piece you seem to write it off as.

Especially Sim City 3000. It's a much bigger piece of software than we're going to write, it's systems are *much* more complex -- and detailed. Our 'game scope' is larger, the entire world versus one city. But the complexity of the interacting systems in Clash will be a mere shadow of what they put into SC3k.

Caesar 3 is really good, for what it is.

* * *

On a side note:

Your question about saving and reloading made me think.

That's the kind of tactic I use when solving a 'puzzle' game or adventure game. I have to get to the next 'point' in the puzzle, so I save to make sure I can continue solving it to the end.

In a 'game' against other 'players', you would never even consider reloading like that.

I think you also are thinking about a 'puzzle' game when you talk about absolutely having to have that tax increase now. If it's a puzzle, and you are trying to get to the next 'level', then it makes me mad too to lose when I've gone so far.

But when playing a game, against other 'players', you would have to admit that it's perfectly fair if they increased their taxes three turns ago to build an army, but you have only a 20% chance of being to raise taxes *this turn*.

[Richard Bruns]

axi: I agree. If we connect the models well, the randomness in riots model will spread throughout the game. That is what I mean by the butterfly effect. My butterfly in the jar analogy is meant to show how the randomness in some parts of sim games does not affect the whole game. I think we can avoid that trap. The riots model is a whole nest of free butterflies, and it should have effects similar to what we see in the real world.

Mark: Good idea. Have you ever played an old game called "Global Conquest"? It let players choose what type of random events they wanted, and I think it worked well. We can let player (and scenario designers) alter the type and magnitude of random events. That should be a good addition to the game.

F_Smith:

quote: Adding 'random' events to a game usually annoys me. I hate it when a 'random' event sets me back in these types of games.

I agree. I hate that too. That is why I have been trying to reduce the randomness and unpredictability in the political games. I would hate it if the 'random' outcome of a bribe prevented me from raising important war funds. That is exactly why I have been trying to improve your good model by adding automatic escalation of foul play. Instead of a 'random' event setting me back, my agents will go to the next step and get the thing passed by harsher means.

Are we thinking about the same game? The Caesar 3 game I am familiar with is just a rehash of SimCity. You are basically a mayor who has to take care of a bunch of ant-like sims. The main difference is that you are directly responsible for building economic structures like clay pits, while in SimCity you just zone things. Is there some other game with the same name?

[Richard Bruns]

I have messed up again, saying big things without explaining what I meant. I need to back up and show what I meant by these two statements:

---
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. . .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.
---

By this I meant giving the player about the same power that historical figures had. IMO players should mot be able to dictate every detail of research, building placement, resource gathering, etc. This gives them too much power. Historically, rulers did not order things like tech research. Tech grew without their supervision and when something big happened as a result of this (like the printing press), the rulers had to react to the disruption caused by the new tech.

When a player is able to manipulate everything, their civ becomes like a machine. Farmers, scientists, and builders are doing exactly what the player orders. The players are imposing their will on every aspect of their civilization. They are controlling every little thing. That is the flaw in "sim" games, which include some excellent games that I enjoy playing. Often, the player's civ or city becomes merely an extension of the player's will. That is fun in a way, but it is not what I want for Clash.

Aside from a bit of crime and unrest, the people in current games don't do anything signifigant on their own. They don't create global corporations that demand player's attention. They don't invent things on their own that cause large societal disruptions. They don't mismanage their farmland and then migrate elsewhere in search of jobs. This is what I mean by reacting. The player, like real life governments, should be forced to deal with surprises created by independent agents. The civ should not be controlled entirely by the player.

---
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.
---

Let me give a few examples here:

In Civ 2, you had to micromanage the farming and resource gathering. You placed all of the little people on the proper squares. Every time the population grew or was about to grow, you had to do it again because the people would do stupid things like going to unimproved grassland when you had a harbor and offshore platform. If you didn't manage this properly, you could lose units the city was supporting or the city would go to unrest because they moved off the trade arrows that gave them luxuries.

You also had to micromanage all of those settlers and engineers. You irrigated, dug mines, made all the little roads, and altered terrain. Instead of leading a country, you spent a lot of time being a foreman ordering little things.

This required a lot of tedious micromanagement. Things were so twitchy that the player constantly had to manage them. Yet despite the nasty little surprises that cropped up everywhere, there were no big changes that demanded big actions. Farmland never went barren. Forests were never depleted. Nothing big and unusual ever happened in the hundreds of turns of micromanagement.

I din't want that in Clash. The farming will go on mostly automatically. People improve the infractructure and choose farmland wisely. The people will usually take care on themselves. The player will not have to give a lot of thought and effort to the farming . . . until the ecology model decides that years of improper farming techniques have turned one of your best agricultural areas into a dust bowl.

The tech system is another example. In Civ 2, you had to micromanage those trade arrows and beakers, build all the libraries, and select the next advancement. But there were never any big changes that required big action (with the possible exception of the Great Wall becoming obsolete).

In Clash, the player will not have to micromanage the tech like that. Tech growth will happen steadily and automatically if you manage the economy and society well. The player will not have to give a lot of thought and effort to the tech system . . . until someone in your civ develops something like the printing press or factory system that causes profound economic and societal changes that must be dealt with.

While I still think that Civ 2 is one of the best computer games built to this date, these things do annoy me. The gaming experience sometimes seems to be hours and hours of the mindless micromanagement tedium that defines the life of a corporate cog. I don't want that in Clash. I want a game that deals with the big events and major problems that have defined human history.

[Lord God Jinnai]

As far as randomness is concerned, does that include disasters because that is the essence of what disasters are, random catestrophic events. Now they won't be quite random as you can't have a hurricane in the artic circle nor an earthquake in a stable area, but generally speaking each event is random, even if its built up over time like earthquakes. When it snaps is randomly determined. I know IRL it has also do with the rock type, the fault type, the speed of plate movements, etc. but i doubt we're gonna worry about most of that here so it'd just haveto be random.

[roquijad]

I agree with Richard in just every single word he wrote.

I like Mark's idea of giving the player the chance of defining the level of randomness. It's clever. However, we have to see that this is not as straight forward as it appears. F_Smith's political system has more random than the "default" negotiation system, but it is another system. What I'm saying here is that F_Smith's system is not "the default system, yet with more random things", but another thing. So, Mark's idea can be useful for a lot of sections of the game, but won't make F_Smith and people like him happy.

It seems to me that the only model where the random vs deterministic issue has sense is in the govt model, and more precisely, in the policy-setting section. So maybe we shouldn't generalize this topic the way we're doing it.


I had an idea. Somehow, F_Smith's political system now appears to me as a short-run political system. It's like the micromanagment side of what the default govt system is. In the mil model, you can play unit by unit every single battle or simply give the overall military strategy. Maybe we should see it like that in the political system. If I want to play the overall political game just with broad decisions over long periods of time (centuries, FE), then I'd choose the default negotiated system. But if I want in any particular time to play politics in their details, then I would push a button and switch to F_Smith's system. How does this sound to you all?

If we agree on that, our main goal would be then to take care of the switching between the two systems for compatibility. In other words, switching to one or the other shouldn't destroy what was achieved previously.

In terms of AI, we should only code for the default system. Going to political micromanagement would be only a (human) players' option and within it, the player cannot tell the AI to make things for him, because it doesn't make much sense to switch to detail-level politics if you're not gonna participate in it.

Players would then have 3 options:
1) Detailed level (F_Smith and all those who really like that level of political game)
2) Global level (default system for those who want control of politics but don't wanna go to specifics)
3) Global in auto-pilot (AI controlled) (for those who don't like politics at all)

[Mark_Everson]

Cool idea Rodrigo. The only issues would be 'docking' the models so that they produce broadly similar results, otherwise there could be problems. They would also need to be somewhat similar or there could be big problems if the player switches the way they want the govt to work in the middle of the game. Guess we need to see what F_Smith thinks.

BTW I locked the old Gov thread. Could you start a new one soon? It doesn't have to start with anything serious, since you can edit any summary in later. Also, please put in a link back to the previous one at the top.

LGJ:

I don't think anyone wants to make disasters deterministic, so don't worry.
[This message has been edited by Mark_Everson (edited September 03, 2000).]

[F_Smith]

Richard:

I agree with every word. I share all those goals.

It's the proposed solution we're discussing.

We should eliminate that kind of micromanagement need by improving the 'unit' AI.


* * *

Rodrigo:

That's exactly how I've coded it already. That's exactly what I've been asking for. The two systems are fully compatible. We could even have different people using different systems within the same game, altho it might be tricky to 'balance' . . .

People who don't want to play politics just don't. People who don't want to play military combat just don't.

And it is also not isolated. Every phase of the game should have a 'general' level and a 'detailed' level of play. I personally don't like random disasters, and would turn those off -- even tho it is 'realistic'. They're not fun, to me, in a game.

But economics, ecology, all these things will need an 'basic' setting and an 'advanced option'.

And the 'advanced' option should be more than just a more detailed sim, it should be a clever 'game' in and of itself. A 'conflict' based set of rules.

* * *

Mark:

They produce exactly the same results.

The only difference is in how they determine those results. The two systems are perfectly interchangable.
[This message has been edited by F_Smith (edited September 04, 2000).]

[Mark_Everson]

F_Smith:

You say the two govt systems produce exactly the same results... I don't know what you mean by this. FE if ruler political power = 51%, it is quite possible for your system to produce 0% as a policy result (ruler wants 0% and gets it automatically), where the default model gives 75% (with the current one using ruler power/2 as a 'correction to what the people want). For the 75% you could have a case where the people want 100% but the ruler corrects by the max 25% to 75%. How are these the same result? They are almost diametrically opposite!

If you mean the two systems both answer the same questions, but give different answers, that isn't the point. That was the problem I was referring to about different results. I meant if they give substantially different answers to the same question there will be trouble in balancing between the two. And also switching between the two.

The upshot is that when switching between systems Enormous political changes or revolutions could occur just because the player selects or deselects the optional system.

[F_Smith]

Mark:

The two systems both output 'policy' values.

Yes, switching between the two would create govt changes -- altho that is the point of the different game option. Likely you would have to select this option at the beginning, and not be allowed to switch during a game.

So the two options are 'balanced' against each other, in that they can both be used by different players to arrive at policy values. As you said, they both answer the same questions. They have the same output -- 'policy' values.

And we can plug in any other 'policy' value determination options, too.

Here's another option I'm going to code in tonight, just because it should be there and is so easy to implement -- The option for the player's policy prefs to simply *be* the govt policy. One option should be to turn all this govt stuff off completely. Some people won't want to deal with the political side at all.

[Richard Bruns]

quote: Every phase of the game should have a 'general' level and a 'detailed' level of play. I personally don't like random disasters, and would turn those off -- even tho it is 'realistic'. They're not fun, to me, in a game. But economics, ecology, all these things will need an 'basic' setting and an 'advanced option'. And the 'advanced' option should be more than just a more detailed sim, it should be a clever 'game' in and of itself. A 'conflict' based set of rules.

I always wanted something like this, but I didn't think it would be practical in many cases. For example, what could we do to make the tech model simpler or more complex? All I can think of is making trees with more or less complexity, while keeping the equations the same.

However, the ecology model can easily be altered like this. It is essentially a collection of processes, so it would be easy to take some out, add more in, or change them. I can think of many possible complexity levels for the ecology model:

0: No ecology modeling. Ideal for battle scenarios.

1: Like Civ 2. Economic inputs depend on terrain. The terrain never changes naturally, but you can alter it. Ideal for slow computers or low bandwith multiplayer.

2: A "bare bones" version of the current model. A few of the processes are used to make the terrain change in simple ways. Ideal for those who don't care about the model.

3: A simplified version of what I am working on. This will probably end up being the default for a normal game.

4: The full model, with all of the stuff I am working on. Your people will want to manipulate the environment on their own, and you will probably wind up in conflict with them if you want to keep the landscape healthy. (As the immortal leader of the civ, you will be the one stuck with with the long term consequences of their short term selfishness.) This option will probably be chosen by about half a dozen people on the planet.

[Mark_Everson]

quote: Originally posted by F_Smith on 09-04-2000 04:08 AM And it is also not isolated. Every phase of the game should have a 'general' level and a 'detailed' level of play. *snip* But economics, ecology, all these things will need an 'basic' setting and an 'advanced option'.

Ummm, This would Double the AI work, which is arguably the hardest part of the whole project. I really prefer having at least most of the simple models just be stripped down versions of the 'micromanagement version'. That way you write one AI for the complicated one, and for the simpler one all you do is change the player interface. The AI then handles all the depth based upon the player's general orders.

For the cases where Only the player can use the alternate, and there is no AI needed, I've got no serious problem with the proposal. But again we may then have balance issues since the detailed model may be advantageous to the standard one, or vice versa.

[F_Smith]

Richard:

Well, they could turn tech advancement off, they could prefer a fully automated tech system they never make any choices in, or they could prefer to actually select tech 'goals'.

And those seem like good levels for the ecology model.

[roquijad]

That implementation for govt setting being exactly equal to ruler's prefs is for sure needed. As an option, it must be there.

Richard: Can you e-mail me the lattest version of the ecology model? roquijad@cec.uchile.cl


F_Smith: I was thinking about giving the player the option to switch between political systems during play. Just like in the military area, you may be in the mood for playing low level detailed combats one day and in the same game in a later time or another day prefer instead to let it in auto-pilot. Doing this with political system would be really good I think. In this case a higher level of compatibility is needed between both systems. After a switch, the political game should evolve in general very much the same way like if the switching never took place. The switch would give you simply more or less detail (and therefore the "right" level of flavor for a given player). So, let me ask it another way: Do you think is possible for your system to recreate, just in a broad overall perspective, the same outcomes/results (policy values) the default system produces? Changes needed in your current system to achieve that would in your opinion be acceptable or would they destroy what you find fun in your system?