Thursday, 21 August 2014

First Law: Override - AI Archetypes

Greetings friends!

So after my short distraction from Tannus and Brooks yesterday I was able to come up with a system that at first seems simple but depending on the end result may be exactly what I'm looking for.

This answer is what I am calling AI Archetypes.

Now considering the over all theme of this game I find it makes sense to refer to NPCs as AI, or Artificial Intelligence, just like you will have in videogames, and as I am trying to create a ruleset that will act similar to a computer in playing out games, that to me seems a logical term to use.

Ok, so what is an Archetype? This is something that was brought to my attention a good while ago when I was still Henching for Wyrd. One of the shop workers, a great guy by the name of David, he was talking to me about my game at the time; MaliQuest. Now, I was looking for a way of adding variety to the bad guys and up the difficulty without making things too awkward and difficult. He told me about (Pathfinder?) Archetypes, adding a trait to the monsters like Brutal or Magical and in doing so this template automatically adjusts their stats.

I loved this idea and while I never used it in either MaliQuest or it's successor; CRYPTS, it remained locked away in my rules vault (that part of your brain that holds onto useless game rules).

Well after years of holding onto this, it appears that now is time to release this from he rules vault and make use of it, however rather than using it as a stat thing, using it as a behaviour system for FL:O.

So; rather than having lots of complicated rules on each character card that details how they operate they now come with between 2 and 3 archetypes.

The first covers their inactive state, of which there are currently three:

Stationary: Does not move at all - super regimented military types may fall under this, or creatures that go into a state of hibernation.

Rotary: Rotates 90 degrees clockwise - think sentry guns/bots

Wandering: Travels in a random direction - most humanoids.

The second archetype relates to Active or Alert states, ie once they have spotted, or been attacked by players, examples of these currently would be:

Brutal: Runs towards the nearest visible player and attacks them with melee attacks.

Ballistic: Moves away from the nearest player (towards cover if possible) and fires with ranged attacks.

Cowardly: Runs directly away from the nearest player. If cornered will attack with either ranged or melee if available.

The third archetype which I have not settled on would be load out. So Sniper for example might have a sniper rifle and light armour and would always start in cover.

Using sniper as an example this would mean you could have:
Stationary/Ballistic/Sniper. These three simple archetypes then say that the sniper will be placed in cover, will not move and when it detects a player, or a player detects it; will shoot it's rifle at the player.


It was pointed out in a previous post by my good friend Will that this and systems like it will create predictable results. You know that said sniper will never move from it's spot and will react a certain way when engaged.

The more I think about this the less I have a problem with it. It means that players will need to problem solve to deal with the issue, however other random factors like wandering AI or Brutal AI will add an X factor to the proceedings. As a wandering AI could over the course of the game go anywhere on the board then even if there is a safety zone free from the sniper's field of vision, in no way is this location guaranteed safe from other AI.

There is one more side to the system that I think leads itself to an unwritten rule more than anything else - having fun!

Yes there is a chance that when it comes to placing the AI someone may think "ah, we have to deal with a stationary/ballistic/sniper, let's have him facing a steel wall in a shitty corner of the board", and that will make the game a lot easier to win, but will that team have a lot of fun? When you play a team strategy game like this is shaping up to be, what fun is there when the strategy is "run right to the end of the map ignoring everyone because they're all placed in a way to deliberately ignore you"? It's like playing a shooter with the age old /noclip cheat enabled.

So what do you think?

Of course one of the great things about AI Archetypes is that there is no reason why expansions can't introduce new Archetypes; so the story expansion that introduces the Triple Max Prison/Mental Asylum world of Fury 161/Crematoria etc who's inmates all share the Unpredictable Archetype, where their behaviour each term is determined by a dice roll!
- This is just an example, doing so for every type of behaviour would really slow things down for the game, but for a single game/story, why not?

And on that note; stay safe and I'll see you Fringeside!

NB: I was going to refer to the place the game is set the Outer Rim rather than the Outer Fringe. Imagine how that sign off would sound?

- Your friendly neighbourhood Doctor Loxley


  1. Hey

    I think you are still going to require a player to control these archetypes:
    - Which cover does it run towards?
    - which player does it target?
    - Which attacks does it use?

    Unless you really limit the actions for every mob, so they only have 1 attack, 1 target, one direction to move you will need some way to make these choices.


    1. These things will be limited, like it will be nearest cover, nearest player, one attack etc

      And there will be player involvement as someone will have to move the models and make certain judgements (which cover is nearest etc) and I'm not sure how to handle this yet, one option for instance would be to have a player turn where every player controls their characters, then an AI turn where one person performs the actions of the AI, however who controls them varies from turn to turn, maybe it moves around the table clockwise, or it's reverse initiative or something, don't know yet)

  2. Just as an aside - are you thinking the fights are going to be between single mobs, small groups of mobs or large groups?

    This will be a factor as you don't want the 'AI' turn to take ages especially if combat is unopposed, but too few mobs and it will be very predictable


    1. It will depend on the mission. On average the mobs will probably be spread out so you are dealing with 1-2 at a time where each humanoid is about as strong if not stronger than a player.

      It will be very pradictable, but that is quickly becoming part of the game. For those sort of missions (infiltration and extermination missions) it will be a strategy game where the mission essentially gives you a puzzle and you have to figure it out.

      For other missions that are more slow build (think Alien or Dead Space) because of how the mission will play out it will be very different.