I am a bit to late with my September report am I not? Well yes I am. Its because it has not been an easy September and October. I had some setbacks with the AI I was coding for Workname Empire. And putting quality above else I had to restart.
As usual i tried to make a to deep AI in the first attempt and it turned out I couldn’t manage to make it work at any reasonable speed. Like just 100 iterations for a small group of maybe 5 units already took like 20 seconds… And if I actually made the AI look into the deeper future (like a round or 3) its speed dropped to completely unacceptable. So I took a big breath and restarted, this time using an approach that will guarantee speed. Where in the previous attempts I used cloned game states of the actual game state I am now using an abstract representation of the game state. Much like the human mind probably ;)
Basically when considering its moves the AI is now in its own little world, a world that is much more simple data-wise than the real game and thus much faster. I think a speed gain of at least up to a 100 times is going to be achieved by this new approach. This will leave room to add some some optional higher AI settings for players who want to wait longer to get better results.
I have made quite some further gameplay improvements to Workname Empire as well. Longer development time has some payoffs. The most interesting thing to share is that the leader model is really starting to get a lot body. The leaders are getting a lot of personality and development opportunities now. There are four classes: Troopers, Technicals, Administrators and Barbarians. Although I am not limiting the player in their character development there is a penalty for learning skills belonging to other classes. The interplay between regular combat units and the leaders is very good and I think it will be one of the strong points of this game. In a sense this game is going where no wargame has gone before. The game will also show the player that larger forces can be overcome by smaller forces if led by able and experienced officers.
Hope to tell you all the AI is working out in my November report. Hoping to have the interface completed by x-mas.
In the meantime Decisive Campaigns 3 is also slowly progressing. I am adding some advanced new reporting functions as we speak as well as the option to take multiple choice decisions when confronted with specific issues. I am working with an as yet secret other developer on this title, if we feel comfortable in opening up to the public we will. So far so good though.