I've noticed in playing around with the minors AI research priotities that the research AI tends to "splurge" on it's researching. It will research 2 air doctrine techs, or 2 aircraft techs, or 2 artillery techs, or sometimes even 2 land doctrines at the same time, even if the minor only has 1 good tech team. Majors do this too - I've seen England research all 4 assembly line techs at the same time, despite only having 2 good teams. The US and Germany likewise only have 2 good assembly line teams (and a few fair ones), but research all 4 at once in 1940.
Would it be possible to program the AI to only research a tech if the best (or best 2, in the case of majors) tech teams were available, and research something else if it isn't? Or would it be easier to program the AI to never research more than 1 tech from the same branch (Industrial, Armor) at the same time? Or more than 2 from the same tree if a major?
Finally, I ocassionally see the AI research '38 techs in '36. Usually these are techs that are prioritized and other research in the field has been completed. But it really slows things down when the UK begins research of basic naval bombers in '36 (tho they might be getting blueprints from somehwhere, for all I know). Some sort of "skip if more than 400 days ahead" test for AI research would be appropriate - unless playtesting says a different number is better.
Would it be possible to program the AI to only research a tech if the best (or best 2, in the case of majors) tech teams were available, and research something else if it isn't? Or would it be easier to program the AI to never research more than 1 tech from the same branch (Industrial, Armor) at the same time? Or more than 2 from the same tree if a major?
Finally, I ocassionally see the AI research '38 techs in '36. Usually these are techs that are prioritized and other research in the field has been completed. But it really slows things down when the UK begins research of basic naval bombers in '36 (tho they might be getting blueprints from somehwhere, for all I know). Some sort of "skip if more than 400 days ahead" test for AI research would be appropriate - unless playtesting says a different number is better.
Upvote
0