Sponsored By

Pacific Conquest's AI Orders System

I found myself experiencing deja vu for a design problem I encountered 20 years ago ...have I not learned from my mistakes?

Timothy Ryan

August 30, 2017

2 Min Read

This morning I find myself experiencing deja vu as I write a programmable finite state machine for my WW2 strategy game. Back in 96 on MechCommander I had written the AI to the point where the Battle Mechs could basically play themselves.

Given initial general orders and rules of engagement a player could just sit back and watch their Mechs perform their pre-programmed actions like an armchair general seeing a battle plan executed well. Some players would get satisfaction out of that, but not most players. We had banked on the battlefield wisdom that no plan survives contact with the enemy and that the player would have to step in and issue tactical commands to compensate and micromanage the battle. Well, that didn't always happen and it just wasn't very fun.

We changed course, leaving the complex self-sufficient AI and complex hierarchy of commands for the enemies and more micro tactical decisions for the player. This put MechCommander more inline with other real-time strategy games.

Fast forward 20 years. My new strategy game, Pacific Conquest, is not realtime strategy. It is turn-based strategy played across the entire Pacific Theater of War over the course of days and weeks with events happening every hour whether you are logged in or not. Your objective is to capture islands, not assault individual units or structures. This puts you more in the role of a situation room admiral rather than a battle group tactical commander. As I set about creating an order interface for what is ostensibly a multiplayer game, I plan on actually allowing players to program conditional orders to execute while they are sleeping or busy at work, something I may put behind a paywall since these are all the same orders you could execute directly if you never slept and checked your event feed regularly.

My plan was that I could include situational protocols and group tactics to the point where I could have sufficient interface for an AI to take over a player's spot should they abandon the game, a common problem for these long-playing games.

Now my dilemma is whether these optional features are making the game too complex for the casual player (probably, for sure some will balk at it) and whether they may feel cheated if they don't take advantage of the advanced order interface that I may put behind a paywall.

Advice would be helpful ... but I will tinker until this is right.

Read more about:

2017Blogs

About the Author(s)

Daily news, dev blogs, and stories from Game Developer straight to your inbox

You May Also Like