GDC + VRDC@GDC 2018 Session Scheduler

View, browse and sort the ever-growing list of sessions by pass type, track, and format. With this Session Scheduler, you can build your schedule in advance and access it during the show via export or with the Mobile App, once live. Sessions do fill up and seating is first come, first serve, so arrive early to sessions that you would like to attend.

Understanding Your Enemy: A Mathematical Approach to Unit Analysis


Leif Walter (Game Designer, Creative Assembly)

Pass Type: All Access, GDC Conference + Summits, GDC Conference - Get your pass now!

Topic: Design, Programming

Format: Session

Vault Recording: TBD

Audience Level: Intermediate

Most strategy games revolve around the interaction of units with each other. Understanding the characteristics and performance of units is key for many steps during the development process: from design, to AI decision-making, and as a supporting layer if combat between sets of units is simulated. Creative Assembly developed a new approach that relies on the mathematical principle of Taylor expansion: instead of trying to identify the accurate performance value of a unit in a specific situation, they linearize its performance around a baseline value within the situational space of their game. Therein, they use game model results to determine modifiers for different situational dimensions of their game (e.g. terrain, units involved) and linearize a unit's contextual potential around its base value by using these modifier values.This lecture will discuss the caveats that come with this approach and investigate its application for game design, AI, and on automatic battle resolution algorithm.


Attendees will gain an understanding of the benefits of integrating the game model into the process of understanding a unit's performance, instead of purely looking at statistics. Depending on the dynamics of a game model, it's inherent impact on unit performance can greatly influence how statistical differences manifest themselves when two units face off with each other.
Furthermore, an argument is made for finding the right balance between accuracy, robustness, and accessibility of data. Often enough, too much effort is put into finding the absolute perfect solution where it is impossible to cover all edge cases while still creating a high maintenance overhead. A case is made for preferring to be 80% correct in 100% of the cases, instead of being 100% correct in 80% of the cases.

Intended Audience

This talk is intended for game designers and developers, primarily for strategy titles where a thorough understanding of unit performance within the respective game space is required. Additionally, designers working on under-the-hood combat resolution mechanics (auto-resolution of battles) as found in many grand strategy games will find this talk very interesting.