Secrets of Gosu: Understanding Physical Combat Skills of Professional Players in First-Person Shooters

Title of the Paper

Secrets of Gosu: Understanding Physical Combat Skills of Professional Players in First-Person Shooters

Paper Information

• Domain: Research on physical combat skills of professional players in first-person shooter (FPS) games
• Keywords: First-person shooter games, esports, performance evaluation, dataset, user study

Research Background and Issues

• Identified Problems or Challenges:

  1. Professional FPS players exhibit significantly better performance than amateur players, but the specific skills or behaviors responsible for this difference remain unclear.
  2. Speculations about professional players' skills in online communities have not been scientifically validated.
  3. Current quantification of FPS player skills often relies on simplistic in-game metrics (e.g., kill/death ratio), failing to uncover hidden skill characteristics.

• Significance:

  1. Understanding the skill mechanisms of professional FPS players is crucial for developing scientific training methods, player trading, and advancing esports as a serious academic discipline.
  2. Scientifically validating specific behaviors' effectiveness in gameplay can help improve amateur players' performance and guide efficient practice.

• Research Motivation and Related Work:

  1. This study was inspired by discussions in online communities, selecting common skill hypotheses for investigation.
  2. An innovative game behavior recording system was developed to collect comprehensive data from professional and amateur players, quantitatively validating these hypotheses.

Solution

• Research Methods:

  1. Hypothesis Construction:
     • Extracted hypotheses from communities, high-view YouTube videos, and professional blogs, selecting 13 hypotheses applicable to general FPS games.
  2. Data Collection:
     • Designed a comprehensive game recording system incorporating six behavioral sensors (motion capture, eye tracking, mouse, keyboard, electromyography, and electrical pulse sensors) along with in-game data recording.
  3. Participant Recruitment:
     • Recruited 8 professional and 8 amateur players, collecting data through 30-minute matches in a one-on-one mode of CS:GO.
  4. Quantitative Performance Metrics:
     • Proposed 15 quantitative metrics covering four categories: aiming, character movement, physical skills, and device usage/settings.

• Innovations:

  1. Introduced data-driven quantitative performance metrics, overcoming the limitations of relying solely on in-game statistical indicators.
  2. Developed a multimodal sensor system ensuring high-precision synchronization of multi-source data.
  3. Systematically validated community-derived game skill hypotheses and made the data publicly available for future research.

• Implementation Steps:

  1. Built the data collection system and verified its validity and accuracy.
  2. Designed experiments and recorded player data, capturing in-game behaviors and event information via dynamic link libraries (DLL).
  3. Preprocessed and analyzed data, validating each hypothesis and conducting statistical significance tests.

Research Findings

• Specific Results:

  1. Out of the 13 hypotheses, 6 were statistically significantly supported:
     • Professional players excel at controlling weapon recoil during aiming (Hypothesis A-3).
     • Professional players adapt better to mouse sensitivity settings (Hypothesis A-4).
     • Professional players integrate crouching movements more frequently during character movement (Hypothesis M-2).
     • Professional players use their arms more for aiming (Hypothesis P-1).
     • Professional players utilize larger mouse pad areas (Hypothesis D-1).
     • Professional players position keyboards at steeper angles (Hypothesis D-2).
  2. Other hypotheses, such as "enemy aim stickiness (A-1)" and "gaze duration for environmental detection (P-3)," were partially supported or contradicted expectations.

• Comparison with Existing Solutions:

  1. Surpassed traditional FPS skill analysis methods based on simple statistics (e.g., kill/death ratio).
  2. Emphasized behavioral process data collection, providing higher credibility for studying FPS game skill characteristics.

• Experimental and Evaluation Results:

  1. Professional players significantly outperformed amateur players in metrics such as "shooting reaction time" and "mouse operation efficiency."
  2. Professional players using low mouse sensitivity settings performed better and relied more on larger mouse pad areas.

• Limitations and Future Directions:

  1. Limitations:
     • Professional players' lack of experience with the experimental game (CS:GO) might affect performance data.
     • Experimental settings differ from real competitions, such as reduced psychological pressure due to small map configurations.
     • Lack of research on team skills in multiplayer environments.
  2. Future Directions:
     • Validate the proposed performance metrics across various FPS games.
     • Investigate the impact of devices and game interfaces on players' combat strategies.
     • Use cognitive experiments to separate the effects of "innate ability" and "acquired training."