Concord Dan Posted April 14, 2006 Share Posted April 14, 2006 After studying the following 1. My 1st game as the Axis at 100%, which I won in Sept 1945. 2. The purchase and research scripts. 3. Posts here of peoples 1st impressions. I have discovered what appear to be 2 problems that need to be addressed regarding the scripting system. 1. The scripts texts (purchasing, research, etc) fire one after another without regard for the amount of MMPs spent by the prior script. IE If the purchasing script spent all of the MMPs then there will be none for the research script or others. 2. The lines within the scripts fire one after another without regard for the current force structure or the prior lines. IE If the purchasing script has an line #HQ= 25 [1], then the purchasing script will buy a single HQ 25% of the time each turn assuming there are available MMPs. Since the order of purchases has HQs at the start of the list it is easy to see how the AI would rarely purchase Tank groups further down in the list. Observations 1. At the end of my game in Sept. 1945, The US and UK were still at 80% industry. They had infantry weapons around 3 and AT around. The US had zero tank groups. The forces were comprised mainly of Corps. How this happened; the purchasing script spent most of the MMPs on the first items in the listing. (Corps are listed right after HQs). Then, on the few turns the purchasing script did not spend all the MMPs, most of the research went into infantry weapons since industry is at the bottom of the research script. So, there was never any research spent on Industry for the AI. 2. Another poster mentioned when playing as the Allies, that the Germans seemed to have more HQs than needed in Russia. How this happened, the Germans have a purchasing script with #HQ=25 [1] for around 12 turns. This will yield around 4 HQs purchased during that time. HQs being 1st in the purchasing script. Suggestions 1. Add a parameter limiting the amount MMPs a script type can use each turn. IE Purchasing script can only spend 50% of available MMPs, research only 25% etc Example #MMPS 50 meaning this script will only spend up 50% of the current MMPs. 2. Add a 3rd parameter to the purchasing lines (Currently, the 2 parameters are % chance of purchase and # to purchase) The 3rd parameter would be a comparison to the total on the map. Example #HQ= 25 [1] [5] this would allow a 25% chance of a single HQ purchase assuming there were not already 5 HQs on the map. Link to comment Share on other sites More sharing options...
Lars Posted April 14, 2006 Share Posted April 14, 2006 The AI always loved buying HQs Think it should read IF (number of land units)/6 > (number of HQ's) THEN 25% chance HQ buy. Of course, that might need to be tweaked a bit to allow for HQ supply linking. But still, the AI should check the number of units on the board first before building HQ. Link to comment Share on other sites More sharing options...
Blashy Posted April 14, 2006 Share Posted April 14, 2006 Wow, nice work on this Concord Dan. Link to comment Share on other sites More sharing options...
Edwin P. Posted April 14, 2006 Share Posted April 14, 2006 Good suggestins and analysis Concord Dan and Lars. No reason to purchase HQ if ratio is > 1:X I would also consider adding a script that controls the order of script activation - i.e. Units/Diplomacy/Research. You may want to invest it all in Diplomacy or Research first, especially, after France surrenders. Link to comment Share on other sites More sharing options...
Liam Posted April 14, 2006 Share Posted April 14, 2006 Well, my honest opinion here is why bother. The AI is worthless anyways, get a PBEM or IP game going and get a thinking man's game going. The AI will never stand up to a human. The human will always find an edge the AI cannot. The only way to make the AI competitive at all is to give it about 5Xs-20Xs the units that you can ever possibly build. More or less what games like Civilization has done. I played the AI the first time in SC2 and I beat it without even sweating. No matter what the setting, I don't believe I'd lose more than few a units to beat it either. It's sort of like playing a 8 year old boy who was little slow in class. At 8 I'd give you a challenge, LOL maybe more like a 6 year old. Link to comment Share on other sites More sharing options...
Blashy Posted April 14, 2006 Share Posted April 14, 2006 Liam, the AI can IMHO be extremely challenging in sC2. But it is a work in progress. I think 3 months from now we will have someone who has modded an extremely challenging AI at the hardest setting. The scripts all you to configure it, so the more you think of what if, you can script it in to cover that weak spot. It will not be as good as a human, but I think it will be challenging enough to possibly make you loose a few against it. Not now, I know. Link to comment Share on other sites More sharing options...
Liam Posted April 14, 2006 Share Posted April 14, 2006 Blashy, you're right if you script an AI to make a move a human isn't expecting and beef up it's strength you can make it a challenge. In Civ, they give the AI per Advancement level a tech increase, production increase, and aggression increase. In General the AI makes many errors but has soooooo many advantages and much unpredictability that you will not always know what to do. You might leave a flank uncovered that gets hit.. This game should evolve that way but AI is definitely not where I'd hope you guys put your R&D. Like any other good wargame that is really made for multiplayer, it's a waste You should put it on advancing the game even further in other areas Link to comment Share on other sites More sharing options...
ASHBERY76 Posted April 14, 2006 Share Posted April 14, 2006 Civ4 has good A.I,so does Galciv2.Even HOI2 doomsday now had good A.I.SC2 with its much simpler gameplay SHOULD HAVE BETTER A.I.I look foward to the work done by the developer in the patch. Link to comment Share on other sites More sharing options...
Concord Dan Posted April 15, 2006 Author Share Posted April 15, 2006 Looking at the scripts I believe that this game will eventually have a GREAT AI. The AI is easy to modify. Once a weakness is found, all players can suggest and implement an improvement. Also, the complexity of the AI looks to be limitless. Link to comment Share on other sites More sharing options...
Blashy Posted April 15, 2006 Share Posted April 15, 2006 You and Edwin should work together, both of you seem to be comfortable with scripting and anxious to work on the AI. I look forward to see what you both do. Link to comment Share on other sites More sharing options...
Recommended Posts