Re: Newsgroup suggestions - for moderators

From: David desJardins <desj@math.berkeley.edu>

Date: Thu, 01 Oct 1998 00:00:00 +0000

Scott R. Turner <srt-nospam@pbm.com> writes:
> To take the dropout example, a very good solution is for the computer
> to take over the dropped position and play it just as a human would.
> However, except for trivial situations, that is very difficult to
> program.

I don't think I agree that that's a "good solution", although it's so
hypothetical in nature that perhaps I just take it differently than you
mean it.

I'm not sure that the typical player would be happy to have one's
peaceful, friendly ally drop out of the game, and have the computer
start running their position as I would, which most likely means
conquering their territory and knocking them out of the game.  If the
computer players played "better" than the humans, so that they ended up
"winning", I'm not sure that would suit most of the paying customers.

I don't think that the most fundamental problem with dropped positions
is simply the difficulty of playing the position "competently" or "well"
(although that is certainly *a* problem).  An even more fundamental
problem is that you can't negotiate with the computer.  It can't
remember that you helped it several turns ago and it might as well take
your side against the evil guy who stabbed you both.  In short, it's an
algorithm, not a person.  Since most people are playing PBM because they
enjoy playing diplomatic games that involve negotiation (otherwise, they
could just play two-player games, or solitaire computer games), this is
a serious shortcoming.

(Now, maybe you really do mean by "play just as a human would" that it's
*also* supposed to conduct diplomacy, read and write email messages in
natural language to its neighbors, etc. etc.  If so, then that's so far
beyond "very difficult to program" as to make the words rather
insufficient to describe the obstacles.)

> I've been annoyed in most of the PBMs I've played at how stupid the
> NPCs are.  I don't think that's because the game designer doesn't know
> how to make his NPCs more intelligent, just that he doesn't know how
> to program that.

Hmm.  I don't think this is true, because I don't think that *anyone*
knows how to make intelligent players of the kind of games we are
talking about, much less the typical game designer.  Actually, I think
the "knowing how" of making competent game players is much more
difficult than the mere programming.

I wonder what you mean by "knowing how" to make a computer player more
intelligent.  I speculate that you must mean something other than having
an actual, specific algorithm which does that in mind, leaving just the
difficulty of translating that algorithm into a correct implementation
thereof.

> (On the other hand, I made the NPCs in Olympia:The Age of Gods a tad
> more intelligent than those in Olympia, and a number of players were
> quite put off.)

Well, I don't have any direct knowledge of that game.  But I would
hypothesize that the players may not have been put off by the
"intelligence" of the NPCs, but by the goals that they pursued with that
intelligence behind them (per the discussion above).  For example, if
you coded "intelligent" NPCs which targeted certain players, carefully
planned their demise, and then launched coordinated surprise attacks on
them, then players might well be put off (while if human players did the
same thing, they would accept it as part of the game, although they
still might be somewhat displeased).  On the other hand, if you coded
"intelligent" NPCs which used their intelligence simply to go about
their own business while avoiding conflict, or to passively but
effectively defend certain key locations (using the same game mechanisms
and tools that are available and known to the players), then I would be
surprised if many players were annoyed by that.

David desJardins

References

Re: Newsgroup suggestions - for moderators
"Scott R. Turner" <srt-nospam@pbm.com>
Mon, 28 Sep 1998 00:00:00 +0000
Re: Newsgroup suggestions - for moderators
David desJardins <desj@math.berkeley.edu>
Wed, 30 Sep 1998 00:00:00 +0000
Re: Newsgroup suggestions - for moderators
"Scott R. Turner" <srt-nospam@pbm.com>
Thu, 01 Oct 1998 00:00:00 +0000
Re: Newsgroup suggestions - for moderators
Carol Mulholland <Flagship@flagship-uk.demon.co.uk>
Thu, 24 Sep 1998 00:00:00 +0000
Re: Newsgroup suggestions - for moderators
lindahl@pbm.com (Greg Lindahl)
Thu, 24 Sep 1998 00:00:00 +0000
Re: Newsgroup suggestions - for moderators
Carol Mulholland <Flagship@flagship-uk.demon.co.uk>
Thu, 24 Sep 1998 00:00:00 +0000
Re: Newsgroup suggestions - for moderators
"Steve Pritchard" <steve@spelbind.demon.co.uk>
Fri, 25 Sep 1998 00:00:00 +0000
Re: Newsgroup suggestions - for moderators
Carol Mulholland <Flagship@flagship-uk.demon.co.uk>
Fri, 25 Sep 1998 00:00:00 +0000

Referenced By

Re: Newsgroup suggestions - for moderators
"Scott R. Turner" <srt-nospam@pbm.com>
Mon, 28 Sep 1998 00:00:00 +0000
Re: Newsgroup suggestions - for moderators
David desJardins <desj@math.berkeley.edu>
Wed, 30 Sep 1998 00:00:00 +0000
Re: Newsgroup suggestions - for moderators
"Scott R. Turner" <srt-nospam@pbm.com>
Thu, 01 Oct 1998 00:00:00 +0000
Re: Newsgroup suggestions - for moderators
Carol Mulholland <Flagship@flagship-uk.demon.co.uk>
Thu, 24 Sep 1998 00:00:00 +0000
Re: Newsgroup suggestions - for moderators
lindahl@pbm.com (Greg Lindahl)
Thu, 24 Sep 1998 00:00:00 +0000
Re: Newsgroup suggestions - for moderators
Carol Mulholland <Flagship@flagship-uk.demon.co.uk>
Thu, 24 Sep 1998 00:00:00 +0000
Re: Newsgroup suggestions - for moderators
"Steve Pritchard" <steve@spelbind.demon.co.uk>
Fri, 25 Sep 1998 00:00:00 +0000
Re: Newsgroup suggestions - for moderators
Carol Mulholland <Flagship@flagship-uk.demon.co.uk>
Fri, 25 Sep 1998 00:00:00 +0000