Announcement

Collapse
No announcement yet.

How to improve the AI's?

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #31
    I didn't want to coment on it, but I have also had much truoble using strategy #9
    What type of idiot quotes himself
    -paramir

    Comment


    • #32
      Originally posted by binTravkin


      I think we could split up the programming AI in a couple of different modules:

      1.Terraform
      2.Governoring
      3.Battle Tacs
      4.Diplomacy
      5.Global management (e.g. research, SE)
      6.(reserved for something I've forgotten)

      Some of those modules will have to interact, but the terraform one is almost a stand alone.
      If we manage to get automation to a level it can create intelligent grids of terraforming, the AI will be so much stronger by that simple thing.
      I agree about the terraforming completely, it seems to me to be the easiest fix, and has a huge impact on the game. I heard that one of the reasons that the AI forming is so bad is that there was a last minute change in some of the production values for certain types of terrain. Unfortunately the AI was not redone (or never got the hang of the changes) and this is why some of the bizarre choices are made. I don't know if this is true or not, but there was a big discussion here about it a few yeara ago. I always shudder when I look at AI bases, as in almost any case more production could be gleaned for less former time.
      He's got the Midas touch.
      But he touched it too much!
      Hey Goldmember, Hey Goldmember!

      Comment


      • #33
        I agree about the terraforming completely, it seems to me to be the easiest fix..
        Yes, it's not only the easiest fix, but the most beneficial, as it seems that the automated formers for a human use the same algorithm as AIs formers.

        We can keep this tradition and make a unified module for both.
        Actually I have been thinking about it and I think I know how should it look like:

        1.There must be a couple of predefined strategies.
        Currently candidates are Forest&Forget, the specialist approach (building Condenders and holes) and the traditional approach - using farms and mines.

        2.There must be a priority table containing all possible commands of terraform.
        For each you can set a "priority" from say 0(not build at all) to 3(build at every square available and form unavailable squares to available if possible).

        If you, say, set the hole priority to 3, it will build holes in almost linear, 1 space grid and lower land/clear fungus were needed to ensure a hole can be built there.

        3.There must be "The Grid Definition Table".
        Of course, you must have a drop-down option which would show you the most usual grids used by human like the:

        cc
        hx

        one (b = condenser, h- hole, x - anything, most probably base, if not then condensor again.)

        And also contain a special display which will interactively show the grid in form of grid of buttons in their respective places. You can then click on any grid item you want and click and change to anything you want.

        Similar grid definition table should be available for bases and they should interact with each other in a way like:

        Terraform grid has:

        cc
        cb

        (c = condensor, b = base).

        Base grid has

        xb
        xx

        (x = anything but base, b = base).

        so the automated former actually understands that

        cc
        cb

        the bolded "c" should be ignored and it should only put road there (and probably a sensor - it should also be definable) to allow CP travel there faster.

        For a human, these interfaces should be available anytime during game, for an AI it should be available in scenario editor for the CMN and if it's not set, AI should "take a look" at the terrain it has and the factional traits it has and "decide" which setup to use and which terraform options to prioritise.
        -- What history has taught us is that people do not learn from history.
        -- Programming today is a race between software engineers striving to build bigger and better idiot-proof programs, and the Universe trying to produce bigger and better idiots. So far, the Universe is winning.

        Comment

        Working...
        X