The Wargamer

Introduction

Like many seasoned grognards, I grew up playing Avalon Hill’s Third Reich back in the 1970’s. This board wargame designed by John Prados and Don Greenwood was a classic and set a benchmark for the World War II grand strategy game genre. In 1992 Bruce Harper updated the game as Advanced Third Reich, and then again in 2003 revamped the whole game as A World at War. John Prados also re-released the original classic in 2001 as John Prados’ Third Reich. The only computer game adaptation was Third Reich PC, a DOS-based game with VGA graphics released by Avalon Hill in 1996.

Third Reich PC was a faithful adaptation of the original board game, but the AI was frustratingly poor. This begs the question—what exactly do solitaire players expect a computer opponent to be capable of? We are still a long way from programming a generic AI to truly think like a human, especially for something as complex as World War II. However, there are a few goals that should be attainable. First, the AI should be able to handle tactical/operational situations involving movements and combats and other game mechanics within a game turn. Second, the AI should be able to make and execute sensible operational/strategic plans over a number of turns. It should be able to select from both historical and ahistorical grand strategies. Third, for variety and replayability there should be variants and other randomness included in the game to require both the human player and computer opponent to react to fresh situations and surprises.

When Strategic Command European Theater was released back in 2002, designer Hubert Cater introduced players to a new computer opponent based on fuzzy logic AI principles. It proved itself to be quite capable on the tactical/operational level, fairly good on most strategic planning, and provided enough randomness to ensure every game was different. The most notable weakness was a lack of strategic naval planning that prevented both the Axis and Allies from reinforcing North Africa and skewed the game by seriously neglecting this theater. A nice game editor was provided that allowed players to make some adjustments to compensate for some AI weaknesses, but not enough.

Strategic Command was not exactly Third Reich reincarnated, rather it was more like Clash of Steel that I had never played, but it was close enough to grab my interest. I became an active playtester for Hubert Cater and helped with the development of the series. Hubert’s design decisions have produced a very fun and popular game series, but the essential seriousness of Third Reich, which Hubert had never played, was not exactly a priority. After getting Strategic Command 2 Blitzkrieg up and running in 2006, I began to shift my attention to developing my own customized campaign from scratch to pick up where Third Reich PC fell short. In this Part I article, I will describe in general terms how the AI has evolved in the Strategic Command 2 series and provide some developer comments about my own Advanced Third Reich campaign. In a Part II article, I will provide an after action review of a hands-off AI versus AI game to demonstrate some of the capabilities of both Axis and Allied AI scripting in Advanced Third Reich.

Overview of Advanced Third Reich

With all due respect to John Prados and Bruce Harper, I chose to call my custom campaign Advanced Third Reich because that is what it is, and more. To avoid any possible conflict of interest, it has never been included in the official Strategic Command 2 releases and has been available as a free download at Combat Mission Mod Database. Basically what I have done as a personal project is taken the updated Third Reich map, country economics, force pools with national characteristics and scenario orders of battle and created a fairly faithful adaptation of the classic board game, for both a 1939 and 1941 campaign. Where Third Reich had quarterly turns, I implemented monthly turns for more operational richness. Where Third Reich had unit stacking in hexes, I halved the map scale to be essentially two tiles for every hex to accommodate no stacking in Strategic Command 2. Game mechanics and some unit types are obviously different, but new features such as fog of war, weather, headquarters bonuses and more enhance the gameplay experience in many ways. In general, though, things in the computer adaptation happen more or less consistently with how they happen in the board game.

A World At War European Map	Advanced Third Reich Map

This article, however, is about AI scripting and not the design notes for my custom campaign; those are included in the download if anyone is interested. But before I discuss how the scripted AI evolved in the game, I want to emphasize how the generic AI and other features already help provide variety and replayability in a game. Besides the fuzzy logic embedded in the AI algorithms, there is an editable +/-1 variability added to combat calculations for uncertainty and surprises in combat results. There is a variable placement option for the setup of minor country forces, ensuring slightly different setups upon declaring war. And there are difficulty settings selectable at the start of each game where human players can give the computer opponent increased production and/or unit experience bonuses to help compensate for AI weaknesses. And let us face facts, there are limitations to what the AI can do and a good human player can easily exploit some reckless behaviors. Some players may call these “cheats” but they are provided simply to help balance a game, and each solitaire player needs to decide for himself where that right balance is.