Jump to content

Options vs Impact from the view of AI

tactical combat turn based challenge AI choices

  • Please log in to reply
No replies to this topic

Moo4 #1 Posted 02 March 2016 - 07:46 PM


  • Players
  • 297
  • Member since:


There is a lot of talk about what combat is missing and that the devs should be adding features x, y and z (mostly from MoO2).  





There's also a strong argument for wanting a player's decisions to matter in combat (with many other posts that I may get around to quoting).


At first view it seems like these two ideas should go hand-in-hand.  The more options available then the more a player's decisions matter, right?


Unfortunately, that is not the case and the critical gap is AI.




Let's consider a very simple action such as retreating from combat.  Retreating effectively is incredibly complex and requires data from both inside the tactical map and (more importantly) at the empire level (fleet production capacity, opportunity costs for the would-be-saved ships, etc).  A common tactic I use in MoO1 on Impossible is to build a single small ship every turn so that the enemy will run their fleet into my 100+ missile bases.  They get the kill and retreat but I take out hundreds of them in the process.  Is the AI 'bugged'?  Not really.  The AI knows it *can* retreat so it starts a fight that it doesn't need to.  The AI destroyed my 'fleet' and therefore accomplished the objective.  Wins the battle, loses the war.


From that perspective I don't want WG focusing on a retreat option.  Instead, I'd rather they focus on an AI that knows when to commit to a fight from a strategic point of view and leave tactical retreating out of it.



The Graph

Retreating is just one example but it hits on the (more important) dynamic between tactical options and impactful decisions when AI is considered.  


Posted Image


Basically, WG should be somewhere along the hypotenuse of the blue triangle.  As their AI gets better the area of the triangle will expand but the reality is that they simply cannot program an effective AI when the action space grows large with the number of options.  MoO2 pretended to be outside the blue triangle but late game combat was fairly limited, predictable, and only seemed interesting because of the cheating AI's ability to produce large fleets.  


As a big MoO1 fan I would like WG to prioritize competitive AI so that my decisions make a difference from combat to combat.  I acknowledge that this leaves a lot of MoO2 combat options on the cutting floor but thats the reality.    




I know there are dozens of these threads already and I intend to participate in them as well.  However, I wanted a place to share some of my thoughts and then link to this thread as needed.  


I have not purchased the game yet.  I will if combat shapes up.  If it doesn't, then I will consider buying the game from the inevitable bargain bin 3 months after release in the hopes that modders will do Steve Barcia some justice.  


Seeing what they have so far I think it is salvageable if they go with stacked ships fighting simultaneous turns implemented in real-time in-between pre-determined pause-points where players can update movement and targets.  




I want a moddable Master of Orion 1.5 with focus on interface and pacing.  Come discuss MoO 1.5 ideas HERE!

1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users