How to improve the AI without the Devs doing too much testing?
-
@Trevan said in How to improve the AI without the Devs doing too much testing?:
@LaFayette , would you prefer having custom AI weights to be written in the map xml? Or as a separate file in the game directory? The weights would control how important a capital, factory, resource, etc should be considered when building out the diffused field. There will also be a diffuse rate that can be controlled.
@Trevan having just done a map with 5 xml variants for you, I could see map makers wanting to customise the AI per xml file, so they can customise each scenario/xml.
If there was one ai.properties file it would effect all xml variants with the same parameters/logic.
-
@Trevan said in How to improve the AI without the Devs doing too much testing?:
@ff03k64 oh, I didn't know about that. How does that file work?
I don't know the details specifically. But it has the name of the xml, or the name of the map as part of the objective. I would guess that you could also have an AI properties file that could be called something like mapname.AIproperties as well.
I believe world_war_ii_global has a single objectives file for multiple xml games.
-
I think it would be best to have a separate file for AI weights. It would make it easier to try different weights.
-
Im now agreeing with ff03k64 and RogerCooper a separate file is best, something like xml-name.AIproperties as this covers all bases, its easier to change and its per xml file.
-
We should be cautious about baking an engine dependency into a map. To continue supporting the map, historically any map properties we have to continue support ad-infinitium. Perhaps such a property file we could just ignore one day, but some caution should be used. I'd also wonder if the weighting values couldn't be computed by the AI, I also wonder if the formula that we are tweaking is truly the right formula and has the right considerations and coefficients to begin with.
FWIW, I think experimenting with a custom compiled game with different formula weights would be a good first data point to know how much value there is in tweaking the values. Once we've gained that experience, we'll be in a better position to judge whether per-map differences are impactful.
-
I would really want allowing map makers to create separate AI files for their games as well.
