General Feedback on Clash

[John-SJ]

Most of you don't know me, but since last weekend I have been in correspondence with Mark, looking for a position on the Clash team. So, since then I have been reading many of the model proposals and the feedback/discussions that follow.

I must say, I am very impressed. You guys have put in a tremendous effort bringing Clash to where it is today!!!

I see that this game is much more finely modeled than any of the past Civ type gamesand this can be a good thing. The *opportunity* for historic accuracy is definitely there to a greater extent than ever before. But as I see it, it also opens the possibility for a real nightmare in debugging and balancing.

In Civ II there are relatively few interactions that are all pretty simple and straightforward. Fo instance, depending on government type you get less or more military units that are free, requiring no support, and the support paid for every unit is the same, 1 shield. Depending on gov't type each square generates more or less trade. And so on. Very limited but reasonably realistic interactions.

In clash I see a gov't model with dozens of interactions, a social model with dozens of interactions, and culteral, military, etc. etc. models, all with dozens of interactions. In order to obtain a (relatively) historically accurate outcome, all of those interactions will need to be at least reasonably realistic as well. From a coding/debug/balancing point of view I fear that this will be a MONUMENTAL task!

I am also concerned about the complexity the player will have to deal with. I am a micromanager, but if I have to be aware of all of the hundreds (thousands?) of intricate interactions in order to play, well, I probably won't play. I think about the detailed tables of data that have been published for Civ II, and the thousands of players who seem to have those tables memorized. It seems likely to me that a set of tables that describe the variables and interactions in Clash would fill a good sized book. Even if such a book were available, no one would want to invest the time to read it.

Finally, are you sure that the system as proposed will run acceptably on any but a cutting edge system? It seems to me that you've got a simulation environment orders of magnitude more complex that any Civ type game to date, and it's be written in an object oriented language with more overhead than C++ AND it's an interpretted language.

John

[Mark_Everson]

Hi John:

Well, let's see, where to start? I'll take your points in order for want of a better scheme... You're completely correct that debugging and balancing of our complicated models will be more difficult than for simple ones. If it should turn out that a model is too complicated to stabilize after a moderate amount of effort, then I think we need to make it simpler. One of the reasons I would like as many people as possible to look over the models is to get input on just this aspect of the models. In addition if a model is too complicated for the AI to comprehend, and perform in a decent manner, then it will also need to be modified. The person in charge of each model will need to be able to provide rules of thumb that will produce successful results so that we can prime the AI with good strategies. If they can't do that for their own models, then the models are obviously too complicated. Personally, I don't think that debugging will be too awfully bad with a decent OO approach. Balance is certainly an open issue. But as I state below, sticking relatively close to history means that a lot of the play balance is already partially taken care of for us.

On complexity of systems, and the lack of it in civ2... Civ2 is great for what it is. If it had decent AI, I would probably still be playing it, and never would have even started Clash. Civ is so straightforward that people try to look for optimal strategies based on the crude world model. But in the civ2 strategy forum we can't even agree on what that optimum might be. So it's simple, but this simplicity doesn't really do you very much good IMO. And that simplicity is very dangerous. At least if you want to connect it with real history. One lesson we learn from Civ is that a country can be at continual war from the dawn of time to the present and succeed marvelously. What Bull! One thing that most of the people in the Clash project want to achieve, is to have a history in the game that behaves a little bit more like real history.

From a micro-managers perspective... The complexity that the player Has to deal with in Clash (if we get the AI right) will be Much Less than that in civ2. The whole idea in Clash is that you should have to check out very few if Any tables to play the game successfully. The idea is to mimic the real world in such are way that if you do something, you will get, roughly, the real-world response. If you tax the people more heavily, without giving them anything back, you will severely reduce economic growth, and piss off the people. So what if we don't get exactly the right amount of suppression of economic growth, or anger/sullenness of the people. Rather than balancing a very arbitrary model like the ones civ uses, so long as we get the direction of the feedbacks right in our more realistic models, something vaguely correct should come out of it. Will it be hard? You bet. And you're right that if it's too hard we will simply have to trim the models.

And what exactly is it that you really want to micro-manage? Does it really excite you to move little pretended people around and tell one to farm this turn and mine the next turn? The civ system forces you to do that to play a good game. But it has nothing to do with the grand strategic challenges that I enjoy most from civ! I want to be able to order my advisers to prepare for war, with very general instructions, and have the preparations done. That's what a Leader would do. A leader would not go to every single city in the empire and say hmmm, let's stop building a library here and instead build a catapult, or should that be a phalanx? I want to say that I want an army of this particular general composition to be at this border at this time and hear "Yes Sir". Or I might be told by an adviser that it is impossible, and have to come up with a different plan for Global Domination . Do you like going to every city every turn to make sure that it is not about to revolt because the population is going to increase by a negligible amount? Clash will clearly not have the capability that is available in civ to calculate the precise size that a city will have six turns from now. So what? IMO that is an accounting game, not a strategy game.

The spirit of Clash is to remove the Requirement for abusive levels of micromanagement. If you want micromanagement, as you note, you can bite off a huge amount of it. But it should just be the micromanagement You Want To Do. You are the player, you get to decide. If you want to run every single individual battle by hand, but have the economy run by an adviser, you can do it. But if you're not interested in anything military, you can put that part on auto pilot too. If you want to stick to really broad strategy, you should be able to play Clash in an evening.

On the type of machine that can acceptably run Clash... I think Clash will be able to run reasonably quickly on moderate systems. For the slower systems the player will simply have to reduce the AI level. The thing I think you have forgotten in your analysis is that Java is threaded. A lot of the world model will be executed while the player sits there and thinks about their actions. That's also when most of the AI processing will happen. In Civ, by comparison, other than the interface, the game uses probably a fraction of a second per turn. We are greedy, we will use all of that time.

Is Clash going to be right for everyone? Obviously not. Just off the top of my head I estimate that at most one in 10 people that buy civ-type games will be interested in Clash. People that want a game that a simple mindless rush can win will not be want Clash. Many others won't too, some for some of the reasons that you state.

One of the reasons I think an open design philosophy is so important for Clash is that we can hear arguments like yours relatively early in the design process. Thanks for addressing the design issues you perceive in a frank and open manner. Now, we'll hopefully hear what other members of the project think about these issues. With our collective judgment we can decide which, if any (or all) of the points you raise require action, and of what sort.

I hope you'll be convinced to stick with the project, but regardless of whether you do or not I appreciate your taking the time to frame and present your arguments, rather than just taking off .

Mark

[This message has been edited by Mark_Everson (edited September 11, 1999).]

[John-SJ]

Mark,

Thank you VERY much for your response. Obviously I have not read everything available concerning this project, but nowhere have I seen so clearly stated the "Vision of the creator". I am beginning to really appreciate what Clash is all about. I also want to thank you for addressing my concerns in your response.

IMHO, I still think it might be worthwhile to take another look at some of the designs that are out there and abstract out some of the complexity that can be without compromising the goals of the overall design.

Finally, as I said before I have been looking over many of the Clash models, and there is one in particular that I find interesting, that is the economic model, and that is where I'd like to make my contribution on the Clash project. I may even suggest a simplification or two.

John

[Mark_Everson]

John:

Glad to have you signed up! I'll put you down as the official Econ coder on the duke list.

On complexity... If there's one think I can state with certainty, none of our models are perfect . I'm looking forward to hearing your suggestions.

Mark

[Kull]

Hi John,

Welcome aboard. I'll leave the long answers to Mark , but here's my take on the complexity issue:

1) "It's easier to take things out than add them in". The general idea is to model the real world as closely as possible. If the result is either a game that's too slow or no fun, we cut until we get it right.

2) "Fun" - Fun to play wins vs. historical every time.

3) "Serious AI" - This is the ultimate challenge. A trustworthy (your advisors) and crafty (your opponents) AI is what will set Clash apart from every other game in the genre. And in order for the AI to be both of those things, it must have plenty of information on which to base it's advice/decisions.

Which brings us back to complexity.

[Paul Crocker]

Kull:

Wow! That's probably the most concise, yet accurate and compelling pitch that I've seen yet for Clash. I'm nominating you for the "Propoganda Duke"