Re: Standard Game

[Eric McDonald <mcdonald at phy dot cmich dot edu>]

Hi Brian,

Brian Dunn wrote:
> A few quick ideas which may make the standard game more challenging:
>
> 1) The default has a big ocean often split in half by a piece of land 
> stretching from one end of the map to the other.  The computer is not 
> very good at making 'canals' with bases, so it could be more useful to 
> default to smaller islands.

I'm not sure that I've ever seen the case where the land has stretched 
entirely from end to the other, but certainly some land masses have been 
quite long.

In such cases, it is probably best if the AI builds two fleets, one for 
one ocean and one for the other. Getting it to attempt the building of a 
canal of bases is probably a bit fancy for the time being....

> 2) The AI tries to be clever by building what it thinks it needs at the 
> moment.  I don't know if it takes into account what could be useful to 
> have later on.  The game designer did add unit types to the game for 
> some reason, though, so a (simple?) improvement for now might be to 
> simply have the AI build some of everything, perhaps in a mix 
> proportional to how much it costs to make it.  This could solve the 
> problem of the AI never making any capital ships.  I've also seen it 
> make only bombers and never any transports, for example.

There are some quirky things going on with unit construction choices. It 
is something I hope to look into before 7.5. The formula for choosing 
what to build is somewhat more sophisticated than just the cost. It 
accounts for things such as offensive worth, defensive worth, worth as a 
transport, etc.... And, it may be that not all these worths are 
calculated in an ideal fashion.

> 3) Transports often go into attack mode without any occupants.  Perhaps 
> transports without much offensive power of their own could be told to 
> wait until they have occupants before going into attack.

This is a known issue. Improving the transport AI is among the highest 
priorities I have in terms of feature requests at the Sourceforge site. 
It will definitely be addressed before 7.5. If no other AI issues are 
resolved before then, I will at the very least deal with the transport 
one, since I believe it to be one of Xconq's sorest spots.

> I may take some time to try to figure out how to do these things, but I 
> figure someone else who is much more familiar with the code may be able 
> to rip them out in no time...

Regarding (1), you may wish to experiment with the map generation 
parameters. These are all contained within the GDL definitions of the 
game. Trying is a good way to learn things.

Regarding (2) and (3), working on the C/C++ sources of Xconq is 
required. I would recommend playing around with GDL, either by creating 
a new game or by modifying an existing one, before getting involved in 
the Xconq source code.

Eric