Announcement

Collapse
No announcement yet.

What the AI really needs!

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

  • What the AI really needs!

    The AI needs help, people....and it needs it badly.

    Yes....we all know that the AI is not bad in a tactical manner. Generally, on Higher difficulties they can keep up with tech, and have an advantage in combat and production.

    However, anyone who's ever developed a chokepoint on the AI, or watched AI unit after AI unit bash into a defensive fortification KNOWS that the AI needs help both strategically and tactically.

    Ideas: First off, each Unit should have a "type" like in Civ:2. Attacker/defender/ship/plane/submarine.

    The main unit insignias would be.
    Attacker
    Defender
    Pillager

    Attacker Units have High Attack figures.
    Defenders have High defense figures.
    Pillagers have High movement.

    Now...the AI should be programed so that
    Attackers like attackers, and so they tend to stay together.
    Attackers like Pillagers, so they tend to stay together.
    Attackers and Pillagers do NOT like Defenders...so they tend to stay away from them.

    Now....Units should also have aggression depending on a number of factors:
    Enemy city nearby: Incrase Aggression
    Enemy Units Nearby: Increase Aggression
    Allied units nearby(support range): Increase Aggression
    Tech level better: Increase aggression
    Veteren Status: Increase aggression
    Damaged: Decrease Aggression
    Alone: Decrease Aggression
    Enemy FORTIFICATIONS nearby: Decrease aggression.
    Enemy is on ANY tile, except Ocean, Plains, Grassland: Decrease aggression
    Enemy is on Ocean, Plains, Grassland: Increase aggression


    AI units would be programmed to try to move towards cities, and AVOID heavy defenses.

    I think this would improve the AI. If they're programmed to avoid defenses....(I.E. My Three Veteren Mobile Infantry, fortified on a Mountain in a fortification)then they will make fewer stupid mistakes, and be more of a threat.

    Any more ideas on fixing the Civ AI?
    I dunno. I think nukes are cool..If you're the only one who has em.

  • #2
    L2F:

    What you propose cannot be done in Civilization II — you have to deal with what you can access via the RULES.TXT file. IOW, you'll just have to fiddle with the numbers and capabilities via RULES.TXT and hope for the best. It's what I've done and, for the most part, I'm not disappointed (one of my improvements has resulted in the AI using cruise missiles on cities and certain land units instead of just ships).

    CYBERAmazon
    "I may not agree with what you have to say, but I'll die defending your right to say it." — Voltaire

    "Wheresoever you go, go with all your heart." — Confucius

    Comment


    • #3
      If the A1 fought better it's production advantage on deity would need cutting back, I think.

      A curious thing I've noticed is that as my game has improved and become more aggressive the A1 finds itself unable to devote so many resources to units - it has to build settlers and some improvements or it just can't keep up. Remembering back I used to face tons of A1 ships. Sometimes it even seemed to get stuck in a loop where it just cranked out dozens and dozens of something - noticeable when it was, say, transports.

      But that hasn't been happening for ages. And I can't remember when I last saw an A1 battleship.

      I suppose it is less satisfying defeating the huge production advantage than it would be beating a more witty opponent. But the game - taking combat with everything else - still has enough challenge to engage me.

      The A1 fights better in AoE I think, but that doesn't make up for the lower immersiveness generally.

      SMAC was a bit of a disappointment, in the end. I'm not sure how bright or not the A1 is at combat because some of the units - the helicopters in particular and marines, just turned out to be too powerful. Once you'd found the right ones to build there was insufficient incentive to go on probing the possibilities of others.

      I like the sound of the outflanking ideas in Civ3 and the suggestion that it will build on the way units can be worked in combinations in the Civ games. Interesting to see what sort of tactical astuteness they can give the A1.

      Comment


      • #4
        quote:

        Originally posted by East Street Trader on 05-14-2001 08:00 AM
        And I can't remember when I last saw an A1 battleship.


        Then you should have a try at ww2 scenario with minor civs (see my thread 'Arriba...').
        In my current game, well over 100 battleships have already been destroyed (not by me though, since I'm hiding far away from the mighty guns).
        Aux bords mystérieux du monde occidental

        Comment


        • #5
          Like2frolick,

          Your anti-aggression when faced with superior force would help the AI avoid the hopeless attack syndrome. However, what the AI really needs is to break out of the whole "examining what each individual unit should do" mindset and start having their units work together in teams.

          quote:

          Originally posted by East Street Trader on 05-14-2001 08:00 AM
          I like the sound of the outflanking ideas in Civ3 and the suggestion that it will build on the way units can be worked in combinations in the Civ games. Interesting to see what sort of tactical astuteness they can give the A1.


          Where are you getting this juicy gossip? Is there an official (or even unofficial) web site giving sneak previews into the workings of Civ3?

          Outflanking is a great idea except that the AI is so bad at getting units to their targets now that they'll need a much superior algorithm if they actually want to flank a target.

          If a lone AI legion attacked my city from the side instead of head-on I wouldn't think I'd been out-flanked. Therefore I assume that this mention of flanking implies that the programmers will actually have the AI units working in groups (instead of each one wandering towards the human individually) which would be a great improvement.

          I think the AI could use a huge amount of help in every area of military strategy. I'm sure that MP players would have the best insight into this since they know what works against humans.

          Avoiding defensive strongholds would be good, but if the human has them well placed, a naval assault is often the only way to "get around" them. I think the barbarians show how very effective a surprise naval assault can be.

          Also a good rule of thumb that shouldn't be too hard to implement would be to always match lumbering catapults with some sort of defender. The AI can currently have it's barb leaders always end up with a military barb so I assume maintaining this sort of stacking is doable.

          It could be good if the AI could seige cities and fortresses. By this I mean first choosing a good defensive terrain square adjacent to the target. Next fortify a good defensive unit there and then build a fortress. Finally, when your "seige tower" square (or multiple squares) have effectively withstood human counter attack for a couple of turns, roll in a bunch of catapults and finish him off next turn.

          There is one thing that I hope wouldn't be too hard to program and which (with the AI's superior production) would eventually be effective even in storming mountain fortresses - build up your attack force before attacking. If a knight couldn't take out the fortress/city/whatever - then don't try it again until you can time two knights to hit it at the same time. Didn't work? Send in four at time. Eventually you'll have a sizable enough horde to take down that pesky human. Human counter attacks against large stacks could get to be a problem, but with enough men coming from enough different sides eventually anything will fall. (Different sides? Maybe that's the "flanking" you were referencing.)

          Comment


          • #6
            If they go to troop stacking, which it sounds like. Programming the AI gets a little easier. The only trick is getting the right mix in the stack, and that's not that hard. A large stack will be able to quickly dispense with those bottlenecks, Unless there is an equal stack defending, which will be a heavy drain on resources.

            So long to the days where 3 good defensive units in a mountain city can hold off the golden horde. With any decent mix, a stack of 12 units, with half being ranged units, will toast them, even if they're behind city walls.
            It will make MP a lot more thrilling as each Player will have multiple 12 stacks cruising the board looking for trouble. If you don't have the equiv in each city. Cities will trade hands often.


            RAH
            It's almost as if all his overconfident, absolutist assertions were spoonfed to him by a trusted website or subreddit. Sheeple
            RIP Tony Bogey & Baron O

            Comment


            • #7
              rah,

              Actually I was envisioning the same combat system as Civ2 has, just with the AI being a little more intelligent. (Having units not work independently, but instead realize they have friends who can help them - a unit with a low def tries to stack with a unit with a higher def, multiple units try to time their attacks to be on the same turn, etc.)

              Civ1: Ut-oh, my exploring horsemen opened a hut full of barbarians! But with no hitpoints to wear down, my horse is at full health after each victory. His luck enables him to survive all 8 attacks! Yea super horse!

              Civ2: My foe has decided to stack three elephants outside the city gates. Well, one well-placed rock from my catapult will knock out all three armies. Yea!

              If an attack in Civ3 only takes out one unit in the stack, the city counterattack (via horse, catapult, crusader, whatever) will not be quite as effective as it is now.

              In Civ2, if a city never counterattacks, you can mosey on up enough troops to take out any number of defenders. However, the prospect of a city counterattack taking out a stack of your men in one fell swoop means you need a bit more planning. Maybe you'll have to move your attackers onto many adjacent squares in hopes that there aren't enough counterattackers to take them all. Maybe there are some good defensive squares that you can fortify a defender in to (hopefully) protect all troops stacked there. Maybe there's some open ground so you can amass your fast riders one square away, then descend all in one turn and wipe them out before the city can counterattack. Maybe you'll have to build an adjacent fortress to protect from the "one shot kills all" phenomena.

              Allowing other units in a stack to survive makes the game lean more towards "the guy with more units wins". (Some might say Civ2 is already like this.) Yes it would make the AI easier to program since most of the options above entail careful moving of units to certain squares which the AI has trouble with. I guess there'd still be stack mix (defensive & offensive units) and terrain factors to consider in unit-level tactics.

              Comment

              Working...
              X