Re: PBM design: give away the formulas? From: bc@lnec.pt (Luis Miguel Sequeira) Date: Wed, 30 Jun 1993 15:08:50 +0000 I know this thread is fated to go on forever, so... here's another contribution to it (BTW, a quite _large_ one...): Let's divide all types of games in two (which is something I quite enjoy): 1) Abstract wargames; 2) Simulations. rob@ccc.govt.nz (Rob McNeur) seems to agrre with me: >After having followed the "Formulae in PB(e)Ms" discussion/argument in >rec.games.pbm over the last week, it seems that there are 2 quite different >requirements for 2 different type of games and these 2 different attitudes >came through strongly in the discussions. However, I disagree with his "definitions". Eg., for Rob, wargaming has to be a blood bath: >Type 1: Wargaming :- The objective is to massacre everyone else in the game, This looks like being too hard on Diplomacy :-). You can win by _not_ destroying everyone else, after all. >Type 2: Roleplaying :- The ojective of the game is to have fun, to learn > and explore the surroundings and to interact with others doing > similiar things. There is no "Winner" and no "Looser". > This includes very few games other than the obvious roleplaying ones > such as xD&D, Amber, Gurps etc. Also includes Olympia, and probably > Monster Island although I haven't played it so aren't sure. Hmm, this is but the wording on "open-ended, role-playing strategy games". However, a simulation can also be a blood bath... --- I will try to point out what, IMHO, is the "core" of both types of games. Type 1 I've deliberately called "abstract" wargames. By "abstract" I simply mean: "the game has its own rules, in order to be playable". Say, Advanced Squad Leader is quite realistic (ie. accounting for weapons jamming and loss of ammo, etc.), but, in fact, ASL is just rules, rules, rules. Thus, the game is described by its own rules. That is, if you want to explain what ASL is, you say something like: "ASL is a game with rule #1, rule #2, rule #3, etc. ad nauseam". So, a "pure" abstract wargame is one that can be perfectly explained by means of its rules (I'm obviously including all formulas/algorithms needed to explain the game in full). Say, chess is one of the best examples of such a "pure" abstract game: there are rules for movement and capturing pieces, and there are rules for winning (or losing) the game. In a "pure" abstract game, in theory at least, you can predict the outcome of the game just by having the rules (...& formulas, & algorithms, & etc.). That is, it's a perfect information world, and a closed one, too. The simplest of those games is tic-tac-toe :-); if you have two players playing it perfectly, there is always a draw. Obviously, if _all_ wargames were of the "pure" type, they wouldn't be fun, as the result would be the same and/or quite predictable... Thus, either an abstract wargame introduces some random element (eg. dice), or player interaction (eg. like Diplomacy: rules have no random effect on the game, but players most certainly have!), or both (which is the norm in the vast majority of the games). However, what happens with pure abstract wargames, is the problem of having a "best" way of playing the game, or, at least, a finite set of possibilities to play it. Chess is great fun because this finite set is quite large :-). Diplomacy's finite set of possibilities is even larger (due to the random effect added by the player interaction) - but everyone knows that starting with Russia is not the same thing than starting with Austria. Other games have their advantages/inconveniences as well; say, in Empire in Arms France has certainly the best army - but the Turks need less points to win. Thus, what happens in those games is that there are advantages and disadvantages on the starting position. Chess, in contrast, has equal starting positions. Obviously, the issue of building cheaper troops in larger numbers or expensive ones is another way of keeping the balance of the game, while providing more choice for the players. Ultimately, even if there is a finite set of possibilities to win the game, this set grows astronomically, as the player's choices increase. On the other hand, knowing _all_ the rules, and _all_ the formulas (algorithms, etc.) is normally provided by table-top wargames, and, in theory at least, a player armes with his pocket spreadsheet can always calculate his best move for each turn, according to the state of the current turn. (Normally, in table-top games, there is just no time to calculate the "best turn", but obviously this isn't true for most PB(e)Ms, with days elapsing between two consecutive turns...) Thus, what I "define" as an abstract wargame is a game with the following characteristics: - All rules, formulas, algorithms, tables, etc. are known; - Given the circunstances, it is always possible to calculate the best move, even if ignoring (or minimizing) the effects of randomness (eg. dice) or player interaction; - The game is closed, that is, you cannot add new rules during the game or take some away. The game _can_ be influenced by player interactions, but what they talk about is the game and its rules (eg. in Empires in Arms you cannot develop nuclear weapons, beacause there are no rules for R&D). --- In contrast to this, we have imperfect information games, which are the true simulations. I suppose that "true" simulations don't really exist (at least, I don't know of any). But the main characteristics are, obviously, a total lack of rules, no known algorithms or tables or formulas, and the absolute impossibility of predicting future events solely based on past ones (you can always make an estimate, ie. a good guess, as we do in the real world...). In theory, simulations should be highly adaptable to new rules or concepts developed by the players, thus two players could play the same game in two absolutely unrelated ways (that is, what the game designers provide is only a "frame" to the game, to make it possible to be implemented. The "frame" is, basically, the theme or the concept of the game, and not much more). Another difference from abstract wargames is the way you play on simulations. On abstract wargames, you play by the rules; there are movement rules, combat rules, etc., and rules on how to win. On a simulation, you play by intuition. Your real-world common sense are the guidelines to playing the game. That is, if you are an economicist (no offense meant :-) ), and you _really_ can't live without your spreadsheet to estimate your best move, you _can_ use it in a simulation to predict (or at least to _try_ to predict) what you ought to trade next. On the other hand, if you are a born strategist, you certainly will know from the info gathered in the game what you can do and what you can't just by looking at your troops. There was once a friend wargamer of mine who defined a simulation the following way: "a simulation is a game where you can give real-world orders and get real-world feedback". What this _doesn't_ mean is that simulations have to be terribly complicated! Almost everything can be simulated using a probability model, and a random number generator. For instance, airport traffic is simulated that way, and you can get quite realistic results with but a few lines of code (if you're interested, I can quote you an example in C++ from a book lying somewhere on my bookshelf at home...). That is, if you want to stage a full-scale war, you don't need to implement weapon jamming factors like ASL has, for each one of your 20,000 men engaged in the battle. You just need a small percentage of the weapons misfiring; or you can even ignore this factor, if you have the rough idea that this factor is not _so_ important. Basically, what a (good, as yet unimplemented, or so far I never saw it) simulation ought to provide is an abstraction level which is realistic. Where to "zoom in" the details, and where to leave just a few probabilities lying around, that's the kind of (very difficult and non-trivial) decision that makes a _good_ simulation. You don't need 5-minute-turns and 5-inches-movement-units. You don't need hit points and armor class for each of your 20,000 men. You don't have to provide orders for each individual member of your retinue (say, "MOVE UNIT James Walsh A LITTLE BIT TO THE LEFT"-type orders). The major problem in building a simulation isn't to cover all the details of the real world, but to know where to STOP. Choose a scale for the game, get the relevant orders to be given at that level, and figure out what the (realistic) feedback to those orders ought to be. Obviously, I gave two extreme examples. Most "simulations", if ever someone tries to write them, would _never_ give so much flexibility as real life... :-) A little more than the usual $.02, Luis Sequeira PS. BTW, I'm part of a small group of simulation enthusiats who are trying to develop a real simulation (the first one? :-) ). The group has had its discussions for the last five years, and we have not yet written a single line of code... :-) :-) _________________________________________________________________________ / / Computer scientists do it byte by byte. _/ _/ _/ _/_/_/ _/ _/_/ _/_/ _/ We don't ask for miracles to get the job _/ _/ _/ _/ _/_/_/ done, we RELY upon them! _/ _/ _/ _/ If the job still isn't done, we'll stick _/_/_/ _/ _/ _/_/_/ with Emacs instead... bc@lnec.pt Luis Miguel Sequeira Laboratorio Nacional de Engenharia Civil Phone 351-1-8482131 Ext. 2752 Centro Informatica/Grupo Sistemas Centrais "Don't call me, I'll call you" Av. Brasil, 101 - 1700 Lisboa, Portugal / _________________________________________________________________________/ Up