Re: Galaxy Cargo, Corners & Diplomacy

From: desj@ccr-p.ida.org (David desJardins)

Date: Tue, 23 Mar 1993 03:40:58 +0000

Brian Sutin <sutin@helios.UCSC.EDU> writes:
> An utter waste of time.  There is an optimal cargo ship design.  It
> does not change.

I don't know what version you are playing.  I don't believe that this is
the case in any version I am familiar with.  I can't see how there can
be a single optimal ratio of drives to cargo bays regardless of your
tech levels, your cargo requirements, and most especially the distance
your ships have to travel?  (Not even considering the version 3.x rules,
where in addition to the speed considerations one has to build different
sizes of ships in order to carry different amounts of cargo
efficiently.)

> My program creates orders
> for hundreds of groups every turn -- it would take me hours to do the
> same thing, and days to get it optimal.

You give orders to over two hundred separate groups of cargo ships every
turn???

I run large empires *and* I build several different types of cargo
ships, which you apparently don't.  And I have never had more than
perhaps 30 or 40 such groups.  I am trying to imagine how one could have
two hundred but I really can't see it.  I guess maybe if you produce
drive tech every turn you could have 20 different tech levels of
transports at each of 10 different planets that you are shipping to and
from simultaneously??

In any case, the only way that I can possibly see that you could have
two hundred groups is if many of them are of very similar ships
traveling the same basic routes, and which can be given similar orders.
So the complexity of writing the orders should not be nearly that great.

My personal experience is that writing the actual orders for my cargo
ships, once I have decided on my development strategy, doesn't take me
all that long.  What takes me a long time is formulating that strategy
and directing my production appropiately to make the strategy work well
several turns down the road.  I know that I take things into account in
that process which can't easily be quantified, much less optimized by
any realistic computer program.

>I said that my program made up orders for CARGO SHIPS.  It does take into
>account the diplomatic state of my neighbors, for that purpose.

No, you did not say that.  You said that your program achieved optimal
growth.  Maximizing growth requires balancing a lot of other factors
than just giving orders to cargo ships.  It requires constructing
appropriate ship designs (although I guess you don't believe that),
balancing production of cargo ships, capital, technology, and warships
(which you have to believe in, I think), choosing appropriate directions
for expansion and development, and a whole host of other interrelated
factors.

For that matter, we haven't even gotten into what "optimal growth"
really means.  It means different things depending on your game
objectives.  For example, if your goal is to maximize your growth over
the next two turns, because you anticipate a conflict, then you might
ship capital to a nearby planet which is already mostly developed.  But
if you want to maximize your growth over the next ten turns, then you
would not ship capital to such a planet, but just let it grow to full
size on its own.

Let's just take this particular case: deciding whether to ship
additional capital to a planet which already has a significant amount of
industry.  This is certainly a decision which frequently has to be made,
and often takes me some time as I consider the consequences.  How can
your program possibly decide this on its own every time without input,
and hope to call the results "optimal"?

Let me make a final note on what I think is the central topic, more than
this debate over the exact mechanics of the game.

It seems to me that precisely the reason that people play Galaxy is
because it is a simulation too complicated to be "solved" in the sense
of producing an optimal strategy.  If one really could produce optimal
strategies, I don't see what the point would be of playing the game.  It
would be, as we both agree, reduced to a purely diplomatic exercise.
And then what would be the point of any of the simulation in the first
place?  One might just as well play Diplomacy, or Nomic.

If it were the case that players were able to formulate essentially
optimal strategies, then I think it would be necessary to make the game
harder or more complicated in order to retain its appeal.

David desJardins

P.S. Are you and Greg Lindahl in some sort of "Galaxy-is-trivial" club?
It's amazing how this discussion reminds me exactly of the ones I had
with him on very nearly the same subject, with the same total lack of
agreement.  At least I can reassure you that there is one person out
there who agrees with you.

References

Re: Galaxy Cargo, Corners & Diplomacy
sutin@ra.UCSC.EDU (Brian Sutin)
Mon, 22 Mar 1993 23:35:07 +0000
Re: Galaxy Cargo, Corners & Diplomacy
desj@ccr-p.ida.org (David desJardins)
Tue, 23 Mar 1993 00:25:10 +0000
Re: Galaxy Cargo, Corners & Diplomacy
sutin@helios.UCSC.EDU (Brian Sutin)
Tue, 23 Mar 1993 01:22:27 +0000

Referenced By

Re: Galaxy Cargo, Corners & Diplomacy
sutin@ra.UCSC.EDU (Brian Sutin)
Mon, 22 Mar 1993 23:35:07 +0000
Re: Galaxy Cargo, Corners & Diplomacy
desj@ccr-p.ida.org (David desJardins)
Tue, 23 Mar 1993 00:25:10 +0000
Re: Galaxy Cargo, Corners & Diplomacy
sutin@helios.UCSC.EDU (Brian Sutin)
Tue, 23 Mar 1993 01:22:27 +0000