Re: PBM design and formulae - a fresh start

From: markiel@callisto.pas.rochester.edu (Andrew Markiel)

Date: Sat, 03 Jul 1993 08:32:51 +0000

In article <gym.741614461@dcs.ed.ac.uk> gym@dcs.ed.ac.uk (Graham Matthews) writes:
>In an effort to take some heat out of this debate I am going to make
>a fresh start on this thread. I am going to spell everything out in
>detail and use some examples to get some specifics.

An excellent plan.

>Ok to the debate at hand.
>
>Assumptions :-
>
>a) we are talking about simulation games.
>b) the type of game could be anything (eg: boardgame, PBM, ...)
>
>My position: games, being simulations of something, should be as
>realistic as possible up to the level of the simulation.

First of all, you are assuming that the game is intended as a
simulation, i.e., as an attempt to model some scenario (real
or imagined). Thus, how close the game describes such a scenario
is important. Thus, to take an arbitrary example, if you were
to consider Galaxy in this context, you could describe it as
a "representation" of outer space ship-to-ship development and
combat, and the rules, etc. would be expected to provide a
"realistic" description of how this would happen.

It's a valid assumption, but I'm not certain how widely applicable
it is. Would most Galaxy players consider the game as a simulation
of combat, or as an abstract game that happens to be presented
within a background of ships, galaxy, etc. I would argue that a
significant number of games (in the opinion of their players) exist
simply as a collection of rules, without being "based" on anything, i.e.
the rules are chosen to be interesting, not to be faithful to any
particular scenario that the designer was imagining when he wrote
the game. (These are certainly the games I would like).

In fact, you've argued before against people bringing skills outside
the game into the game, but it seems that here you are implying that
people are bringing in from the outside something very important,
namely a preconceived idea of how things should work (faithful to
the appropriate scenario in question).

>Players
>should be told the general rules of play, the scenario etc, but beyond
>that they should only be allowed to learn things about the game and
>how to play etc through the actions of the entity they represent in
>the game.

In other words, the game should be a black box, the only things
you know about how the black box works are what you can deduce is
reasonable from knowing the scenario involved. You can apply
certain inputs to the black box (defined by what this entity
you represent could do) to get certain output. The input you
choose is initially based only on what you can guess from your
knowledge of the scenario, but is later supplemented by knowing
what output you got from previous input.

Again, you can choose to play this way, but I probably wouldn't
like it. First of all, I'd prefer to know how the black box works,
so that I know what output would be produced by my input (subject
to random factors, limited knowledge, a game state which is
being altered by the other players simultaneously, and a fundamental
inability to perform a complete closed analysis of all possible cases).
Also, some consequences of this are (1) people's initial performance
is determined by how well their perception of the scenario matches
the designer's perception, which is a somewhat random variable.
And (2) if the players don't know how the black box works, then
there is no easy way to (a) identify bugs or (b) identify "features"
that are non-intuitive, but included by the designer. Thus the
first person to find a significant bug/feature has a big advantage
over the others (ideally, of course, there would be no bugs or
non-intuitive features, but such a perfect game seems unlikely).

>As a corrolary therefore I do not think exact formulas
>should be published. The reasons I take this position are twofold :-
>
>a) it adds to the realism of the game. In the real world people
>don't know exact equations for things and yet seem to do ok.
>A simulation should therefore capture this imprecision - if it does
>not, then it is not a good simulation.(even the formulas themselves
>are only approximations to the reality!).
>
Acceptable, subject to the assumptions you have made above.

>b) it makes the "playing field" as level as possible. People who
>are not math or CS majors are not disadvantaged too badly by not
>understanding the formulas. Moreover everyone has the same basic
>learning mechanism, namely the actions of the entity in the game
>that they represent.
>
Again acceptable. But note that formulas does not necessarily imply
a need for math or CS background: anyone with a high school education
can do simple algebra, for example. Diplomacy is a game with formulae
simple enough for anyone to understand. Similarly, in Game 4 of
Star Empires, a 15 minute comparison of firepower vs. cost was
sufficient to show that the destroyer was the most optimal of
the warships. There's nothing intrinsically wrong with designing a
game in which the players are expected to perform some level of
calculations; if you don't want to do those calculations, then
obviously the game is not for you, but that does not necessarily
imply a flaw in the game design.

>Now there are caveats in this position, namely the caveats of :-
>
>a) experience. The more players play the more experienced they get
>and hence the better they get. This is inevitable - publishing
>the formulas would not solve this problem because the non math
>people would not be helped by knowing the formulas for example.

Unless the formulas are simple enough for all the players to
understand. Then all players, new and old, have the same ability
to figure out what they need to know.

>The only way to counter this is to let the players treat experience
>equitably by helping newbies or by playing with people of the same
>experience.
>
One way of helping newbies is telling them the formulas :-)

>b) statistical analysis. Again there does not seem to be all that
>much that can be done about this. Sure lots of statistical analysis
>will help you gain approximations to the formulas that are being used.
>But sufficiently complex formulas, random walk, and many aspects of
>the game that don't depend on numbers (eg. alliance making), together
>with sensible relationships between things (eg. food growth depending
>on season and not on the colour of the king's hair) will tend to
>reduce the advantages of the statistical analysis.
>
Other (more qualified) people have already addressed this point.

>Now to an example. I will stick to one example for the moment.
>We are to simulate a game of tennis. Now one immediately notes
>three things :-
>
>a) the ball obeys some equations, parameterized by court speed,
>air pressure, spin, racquet speed etc.
>b) practically no players know the equations and yet still manage
>to play somehow.
>c) knowing the eqautions would not help a player play better because
>by the time he/she had solved the equations the ball would have
>gone past them.
>
>So if we are to write a good simulation of tennis we should not
>make it equation based, or rather we might have the ball obey
>equations, but the interface with the human would not be through
>equations. We would not ask tell the player - oh the ball is coming
>at you with the following parameters ... what parameters would you
>like to reply with. What we would do instead is say have a small
>picture of the racquet showing its tilt etc (no figures just the
>picture). Through practice the players would get to know how to
>position the picture to hit certain shots.
>
>Now you will note that this is imprecise. The players don't know
>exaclty the formulas for the ball motion, and even if they did it
>would not help them as they don't interact with the game through
>figures. But its also realistic!
>

This is a perfectly reasonable description of a tennis simulation,
and I have to agree that in this case (once we assume that the
game has no bugs, and that the ball will respond in the same way
as a real tennis ball) that there is no need for formulas. (Also,
this is subject to my points above, namely that good tennis players
in real life will have a significant advantage, and you're in big
trouble if the game turns out to be inaccurate in relation to
tennis in a small but important detail, which of course can only
be determined by trial and error since the formulas are not
available).

However, this follows directly from your construction of this as
a scenario which all players are able to envision without extensive
description, and in which players are expected to start from basic
principles and develop skill at the game. If, instead, the game
involved consideration of the fuel requirements of a battlecruiser
moving at 1.7 times the speed of light through a region containing
approximately 60% ionized dilithium, I think you would agree that a
formula may well be needed, since there is certainly no intuitive way
for a player to approximate this, while an experienced player would
know the appropriate value. Furthermore, I doubt it would be interesting
for people to have to deduce this by trial and error; it's more
appropriate simply to provide the formula, so that people can calculate
the movement capabilities of their ship.

Note that this only tells them what they can do, without giving any
information on what they should do. It still remains for the player
to decide how to distribute his fuel supplies to his ships, perhaps
taking into account whether an enemy attack is anticipated, or whether
it is worth the expenditure of fuel to get the ship to the appropriate
place this turn, or wait and arrive next turn.

These types of decisions are what I play most games for; knowing exactly
how much fuel I need to do action X is only helpful in comparison with
the importance of action X, which is a completely subjective quantity
which no formula can provide. It seems irrelevant in this case whether
I have an approximate or exact value for the fuel cost; what matters is
that I have an appropriate value to use in evaluating action X, which
needs to be provided via the rules since I can't estimate it myself.
Thus, providing the exact formula is just an easy way of giving the
appropriate rules; why bother trying to make up rules that are vague
but allow an appropriate estimation of the value, when you can just
give the formula itself. In a properly designed game, knowing the
exact value should not confer any significant advantage over having
an approximate value estimated from the rules.

I guess my opinion is that your conclusions follow reasonably well
from your assumptions, but your assumptions are sufficiently strict
that they exclude most of the games that I would be interested in
playing, so I can't really agree with your argument in the general case.

>I will await some comment before moving onto another example.
>
>graham
>
>PS: when you agree with something please say so as it makes the
>ground a little firmer for both of us.
>--
>Graham Matthews                   Go west, go west
>Dept Comp Sci, Univ Edinburgh     Where there's work so I'm told
>gym@dcs.ed.ac.uk                  California's never cold

-Andrew
markiel@callisto.pas.rochester.edu

References

PBM design and formulae - a summary
rob@ccc.govt.nz (Rob McNeur)
Mon, 28 Jun 1993 09:53:02 +0000
Re: PBM design and formulae - a summary
gym@dcs.ed.ac.uk (Graham Matthews)
Thu, 01 Jul 1993 11:40:27 +0000
PBM design and formulae - a fresh start
gym@dcs.ed.ac.uk (Graham Matthews)
Fri, 02 Jul 1993 12:01:01 +0000

Referenced By

PBM design and formulae - a summary
rob@ccc.govt.nz (Rob McNeur)
Mon, 28 Jun 1993 09:53:02 +0000
Re: PBM design and formulae - a summary
gym@dcs.ed.ac.uk (Graham Matthews)
Thu, 01 Jul 1993 11:40:27 +0000
PBM design and formulae - a fresh start
gym@dcs.ed.ac.uk (Graham Matthews)
Fri, 02 Jul 1993 12:01:01 +0000