|
EricLarsen -> Improving the AI (10/23/2003 10:20:21 PM)
|
If there's one thing that's become rather evident to me it's the way the scoring is done and how it affects AI play. The object of SUDG seems to be to emulate Bill Gates and amass the largest fortune. The scoring is skewed far too much towards valuing money as the most important factor determining victory. I never saw a t-shirt that says "he who dies with the most money wins" but I have seen t-shirts that say "he who dies with the most toys wins".
SUDG should reward players for producing stuff not hoarding cash. I've noticed that when I have a huge stash of cash and convert it to produced goods on worlds or new ships that my score goes down not up. Geez, how republican can you get? Converting cash into finished goods should increase the score, not decrease it. Finished goods should have a multiple value over the cost you paid to produce them so that they're worth several times the same amount of money for scoring purposes. Things that are from newer tech ages should be valued higher than their equivalents from an earlier age. Certain items should be more valuable than others like replicators and research facilities being more valuable than deep mining facilities. I noticed multiple deep mining facilities on AI worlds but not multiple replicator facilities or research centers.
Because of this incorrect skewing of scoring by making cash the most important commodity (how many worlds have you ever taken with just cash?) the AI is skewed towards accumulating wealth and not spending it wisely. If scoring is fixed so that finished goods are the way to score the most points, and improving stuff ups the score then the AI could be made to play better. I've always seen AI homeworlds stuck with type 1 space docks even though their other colony worlds have type 2 or 3 spacedocks. Because scoring says keep your money in your wallet the poor AI does that instead of improving important facilities like spacedocks. I've also seen where better spacedocks are important in improving ship repair turnaround time so that they don't hold up the freighters and waste money. Because scoring says accumulate cash the AI does that rather than build more and better ships. After the initial rush to build up to 5 or 6 ships the AI seems to stop building ships unless you blow them up.
The AI still sucks at attacking worlds, atleast human ones. I see it does a credible job blowing up other AI worlds but that must be because the AI defends worlds so poorly. I'd say that maybe it's better to not have the AI attack human worlds since there's always so much lead time between the first warning and the actual attack. The AI seems to dork around honoring captain's upgrade requests while assembling the attack force. By the time they get around to attack I've always got a sound defense waiting for them. Unless the AI can be made to vastly improve world attacks by making them ignore captain's upgrade requests so that world attacks go in much quicker after the AI decides to do the deed then world attacks against human worlds shouldn't be undertaken unless the AI has a very clear 2-1 or 3-1 advantage in ships and sends them to attack faster than currently is being done.
Random events are not random, they are still balancing events. While I saw one or two good ones early on in games I never saw good random events when I was doing well. When I was doing well then all I got was whacked by bad random events. I won't use random events again.
Pirates are another way the AI wastes money. I've decided not to use pirates or to allow opponents to aid pirates as the AI just wastes money aiding pirates rather than building up its own fleet. Challenging pirates really help me train my crews more than anything else so now I won't bother with that either so that pirates become the small sideshow they should be and not the main game as they are when the AI aids them and they are challenging and numerous.
Eric Larsen
|
|
|
|
