On Wed, Dec 12, 2001 at 08:17:45AM -0800, Tom Holub wrote:
> On Wed, Dec 12, 2001 at 04:11:13AM -0500, Mark Mielke wrote:
> > On Tue, Dec 11, 2001 at 10:13:49PM -0800, Tom Holub wrote:
> > > On Tue, Dec 11, 2001 at 07:33:32PM -0500, Mark Mielke wrote:
> > > > Tom thinks it isn't possible, or that it won't be effective.
> > > Wrong.  I think you're an idiot.  It is certainly possible to create a
> > > netrek robot that could beat the best human team.  It's a very difficult
> > > problem, and not one anyone's going to solve by concentrating on tactical
> > > improvements.  (It's not one you personally are going to solve by any
> > > methods).
> > Tactical improvements are the butter between the bread.
> > If the robot can't fight, it can't get a kill. If it can't get a kill,
> > it can't take a planet.
> The robot code written in 1986 is good enough to get kills.

It is my understanding that this code cheats. Ever since they added
the ability for a robot to run out of fuel, the robots have sucked
quite a bit more. Things like robots having access to "straight"
torpedoes, that don't wobble, I definately consider to be cheating.

You suggest that the Netrek problem space is complex. I can't argue
with this, as I agree.

This doesn't mean that you focus on strategy, and neglect combat. The
entire problem space must be analyzed. As an example, if the robots do
not do an extremely good job under low fuel conditions, including such
things as *not* running, how will they ever deal with LPS against the
other team? Some might suggest that it is not necessary, as INL games
have a timer. As long as more planets are held for the period of time,
they win. The thing is -- I don't particularly care about the INL
component. Beating good humans just means that the thing is working.
They are guinea pigs more than anything. If it can't beat them, it hasn't
reached its full potential.

The true intrigue is in the methods, and in the consistency. Will the
'bot team function well under all circumstances? All circumstances
definately includes combat. A team that can dog-fight incredibly well,
but is not as capable in the strategy department, still have an
excellent chance of winning against a team that cannot dog-fight very
well, but are quite profficient in terms of strategy. It's the whole
analogy of a boxer getting into fight with a martial arts expert. Who
will win? It is really definately hard to say.