Some Thoughts on the AI

Discussion in 'General Discussion' started by Ozzcams Razor, Feb 26, 2019.

  1. Ozzcams Razor

    Ozzcams Razor Master Chief

    First, congratulations to the developers for producing a spectacular game. I have greatly enjoyed the campaign as well as the arcade mode, multiplayer skirmish and the ability to design your own maps and scenarios. Wargroove fills a major gap for me which is turns based strategy gaming on a portable platform – in this case, the Switch. I have spent countless hours playing the Advance Wars series, AOE, Civilization Revolution, Endwar, Panzar Tactics, etc. on Nintendo’s handheld system and like many of the members here on the forum, I was anxiously awaiting the arrival of this game. Wargroove does not disappoint and the vast amount of content was worth the wait. I tried playing Tiny Metal but could not get immersed in that game because it was cumbersome and clunky among other flaws.

    My only criticism of Wargroove is the relatively weak AI. The AI has some major gaps mostly in terms of its defense coupled with at times, an overly aggressive offense. As an example, the AI will produce Trebuchets which can be a devastating weapon. However, it will commonly move it within attacking range of my units and I can quickly and easily destroy them which is a significant loss for the AI. This is true of other expensive units such as Emberwings. Another flaw in the AI is the unusually high focus on taking a town (economy crippling) versus destroying a weak enemy nearby. These AI flaws can be easily exploited which is really the main strategy for beating the game.

    In order to compensate for the weak AI, the game stacks the deck against you in the campaign and in hard modes. Many TBS games use a similar approach but I find that Wargroove relies on a dipropionate amount on creating an advantage through it starting position with the number of towns, units and/or spawning capabilities. Don’t get me wrong, I like the challenges of the campaign and enjoy playing the Arcade mode at the hard level but I would even enjoy the game more and achieve greater satisfaction with a more level playing field and a stronger AI. Programming a challenging and competent AI is not easy and I can appreciate the work that would go into creating a strong AI. While even the Advance War’s AI was flawed, I find that it’s much more competent and competitive than what is found in Wargroove. The AI did greatly improve through each iteration of AW so I am hoping that Wargroove follows a similar path.

    I don’t want this post to detract from the incredible job the developers did with the overall final game which again has been an amazing release.

    So, what are other people’s thoughts on this area?
     
      Axe Garian likes this.
    • Midnight Tea

      Midnight Tea Phantasmal Quasar

      I think the developers said in a blog post that the AI not being perfect is sort of by design. In a lot of ways the AI is meant to be something to cut your teeth on, something that'll only really challenge you if you're a beginner. I do think you're right that the later iterations of AW's AI are in fact much smarter than Wargroove's. Them getting really aggro at wagons is a blast from the past in particular. It takes a lot of work to make a good AI, like you said, possibly even beyond Chucklefish's resources at the moment. I'm of the opinion that the AI we have is alright and all it really needs is to respect fog of war rules.

      Normally I'm someone who balks when someone says "if you still need a challenge, get a human opponent". This is because I have intense social anxieties regarding matches with human beings and especially playing with strangers. I intend to make my own post about this in a bit, but Wargroove's multiplayer is the first stress free game I can even remember. The way multiplayer works does a superb job of getting me challenging human opponents and not leaving me feeling like a total wreck of anxiety. And unlike in Advance Wars, quick play doesn't guarantee I'll be fighting someone abusing a horrendously broken CO like Tabitha. Right now it'll be Nuru 9 times out of 10 but she's downright fair by comparison to the broken COs of the old days.

      But yeah, the way this game makes it very easy to find a challenging human opponent and makes it stress free in doing so really more than makes up for any shortcomings on the AI. I'll talk about my personal anxiety problems in a separate post soon, it doesn't deserve to be an aside. Chucklefish needs to know what a wonderful thing they've done for people like me in regards to multiplayer.
       
        Axe Garian likes this.
      • Hattori_Hanzo

        Hattori_Hanzo Aquatic Astronaut

        The AI has a weird proclivity for building ballistas on maps where it's not very beneficial, even maps where there are no air towers or air units at all. Sure. they can attack ground units and have a better travel rate than trebuchets but to me it's a waste of gold, they are not efficient when no air is around.

        It's also very easy to lure their high price units and even commanders into a trap, basically they will always take the opportunity to attack your CO if they get it. I love to get some cheap units near my CO and leave them exposed to attack by the AI CO. They will move in, then I shut the door by cutting off their escape, heal my CO (or even retreat them), then just pound the enemy CO because they can't move. If you're attacking on 3-4 sides, even cheap units will wear them down. And lower priced units will often just sacrifice themselves on your CO where they are at a disadvantage like being stretched away from their healing and other units. I always say "thanks!" when this happens.

        And yes, as mentioned above, they put way too high a priority on capturing vacant towns.
         
        • Miinow

          Miinow Void-Bound Voyager

          I noticed the same when grinding through arcade runs. AI wants to build 1 of each type of unit, which I sorta understand. This way it creates a versatile army, which doesn't atleast have any major weaknessess, even though the composition is surely sub-optimal. Ballista is like the 1# ground priority, then archer. If one of these dies, the AI immediately wants to build a new one. On sea it wants to build harpoon ship before merfolk. Dragon is probably the overall 1# priority. Great way to destroy AI:s eco is to attract the dragon to do 3 hp of damage to your co and then immediately demolish it with mages. Even better is to camp his tower :D

          To be more effective, the AI should build more capturing units earlygame, both land and sea, and aim to capture vacant cities at least on his side of map. It should also be more reactive on builds. Ballista for example is a hard counter on heavy air - not a opening build :facepalm:

          As for vacant towns, the AI is surely active to capture them, if on sight. However, it cannot generate even a 2 turn plan to capture cities away from his path.
           

          Share This Page