The Air-Speed Velocity of Unladen Gaming
Now I had some ideas on this issue. Id be curious to hear
your views on them.
1) On-line Karma. The idea here is other players could rate
you as a player. Not in terms of how good you are, but in terms
of how well you behave. The better their karma is, the more it affects
yours.
There are challenges to such a system. Firstly, if you beat someone,
he may be a jerk and rate your karma down out of spite. Worse, they
may have friends that would do the same. But as long as your online
network had only one user ID per person (perhaps they could tie
it to their CD serial), this would be somewhat limited. Obviously
if most people playing the game online are jerks, theres nothing
that can be done.
There is a way around this we use on our skinning community website,
WinCustomize. The Admins
turns a considerable number of trustworthy users into super
users. Think of them as having super-karma. These active members
serve as avatars, people who the admins feel are representative
of how they hope the community of users will behave. These avatars
in turn could provide a good basis for giving other players high
or low karma. Although the jerk user and his friends might be able
to harm you slightly, the theory is that the avatars and the hundreds
of people they have promoted, and the thousands of users those people
then promote, would have far more effect. The odds are that you
will play one of these trusted users, the ones who will
have the most effect. Then as a user, you can avoid players who
have bad karma.
2) Synthesized Multiplayer Gaming. The idea here is to have
your AI so good that it plays as if youre online. This is
something were looking at doing in Galactic Civilizations,
which is coming next year.
The idea is fairly straight forward. When you play the game, you
can play it online on Stardock.net. But you play against
the AI instead of against other people. Your score and games are
recorded and put into the Metaverse, a vast network
database. You can visit the Metaverse as a website to see how you're
doing. A galactic map is displayed showing your position. You can
even read an account of your glorious wars that includes any major
events that occurred.
There is a good side benefit. By recording the events, the game
can see what strategies are the most effective and have the AI use
them. What technologies are best to research and in what order?
What ships are most effective and in what quantities? Is winning
through outright warfare better than becoming an economic super
power or vice versa? What is the best tax rate to have? What are
the best spending ratios to have? What are the best players doing?
This isnt anything fancy like a neural net; its just
statistical analysis. The AI can simply download these stats and
use them as its own strategy.
So when you play against the computer, you can choose a difficulty
level that matches your skill level. Itll be as if youre
playing against a polite, infinitely patient human player of the
skill level of your choice. Playing it on easy level? What do other
novice players tend to do? Playing it on toughest level? What do
experts tend to do? The system could easily be expanded to where
users who give us permission could be provided as synthetic opponents.
So you could play a simulated version of me or your friend down
the street. The game could simply use the strategies that player
used when they played online, doing the same basic things (it wouldnt
be exact, of course, but it would be remarkably similar).
Whats particularly cool in the case of Galactic Civilizations
is that it has the concepts of good and evil as a game mechanic.
Events occur that allow the player to choose various options on
how to handle them. Their choices slowly determines whether a civilization
is good or evil. So if Bob the expert player
is always evil when he plays, when you play simulated Bob, hes
going to be evil, too. His simulated ambassador's dialogue will
be based on how Bob answered the same situations when he played.
In this case, the game plays like its multiplayer but is
actually single player. Its not the same as playing friends
at a LAN party, but for those of us who play anonymous strangers
online, this is certainly preferable. I play online because game
AI tends not to be challenging enough. If we can solve that problem,
then direct multiplayer become less of a requested feature.
And in our case, the other benefits ladders, scores, galactic
empire maps, etc. are all available. We hope to have the galactic
map done by the November beta of GalCiv.
Eventually, players will actually see their empire on a map growing
or shrinking as they play.
Conclusions
Any survey done on multiplayer strategy gaming will show that most
people prefer good single player AI to multiplayer features. Its
not that they dont like multiplayer per se. They just dont
like the hassle that comes with it, such as having some vile 15-year-old
swearing at you. For most people, a good single player AI will do
the trick. And for those like me who do love multiplayer, a karma
system would help. Even a flawed karma system would provide an incentive
for people to try to be a little
nicer.
What do you think?
Brad Wardell was the designer of Entrepreneur and The Corporate
Machine at Stardock. He is
currently the Project Manager of Galactic Civilizations, which is
expected to be released in the Fall of 2002. His columns are archived
here.
|