Parallel ranking of 1v1 computer games.

By Erlang Central | Published: March 2, 2012

Glicko-2 is a rating system. Much like its cousin ELO, it rates players in 1v1 games giving each player a score – a belief in the skill of the player. But unlike ELO it is more computationally intensive and arguably provides better results. We show such a system built in Erlang for ranking duels in the computer game Quake Live. And we rank them in parallel utilizing all CPU power available. A novel approach makes use of Ulf Wigers “JOBS” framework to handle the parallelism and data gathering tasks in the system. We also show how to plot such data through the statistical tool R and the ggplot2 package.

Download Presentation Download


  • Jesper Louis Andersen

    Creator of the eTorrent project
    Erlang Solutions

    Jesper is a Danish programming language geek who is now heading up the Erlang Solutions Copenhagen office. Jesper has programmed in numerous different programming languages. He has a keen interest in weaving functional programming with parallelism and concurrency. He likes to try out new ideas from theoretic research by finding a real-world application and building a system around the idea in order to evaluate its usefulness. In the process he likes to apply knowledge from different areas of mathematics and computer science and he has a curiosity for anything new.

    He is the principal programmer and leader of two open source projects, implementing the BitTorrent Peer-to-peer content distribution protocol in Haskell and Erlang respectively.

    Jesper Louis Andersen

Follow Erlang Central:

Have an Erlang Question?

Reach out to the Erlang community