Thursday, February 12, 2009

Project Horseshoe: Multiplayer Game Atoms

The 2008 Project Horseshoe reports are up! We wrote about how to diagram multiplayer games using skill atoms. Truly a brilliant weekend. The discussion was quite wide ranging and as a result the write up became a bit...long. However, the results should spark a few brain cells. Let me know what you think! :-)

Best wishes,

PS: There are some great reports up this year so be sure to browse around a bit.


  1. Hi Danc,

    I am in the process of trying to use skill atoms to figure out a way of improving my entry for the global game jam. I'm finding that its forcing me to think in a very structured manner, rather than just tinkering with the code and adding/removing a rule in an ad-hoc manner. I'm still in the process of listing out the various skill atoms in the game, and the skill chain.

    If I do end up making something that is substantially more enjoyable than what is now, I'll definitely try and to a 'before' vs 'after' writeup showing the problem areas in the skill chain and how they were rectified.

    I didn't find any mention of this in the writeup, but it would seem to me that explicit skills are generally result in more pleasure than implicit skills. To take extreme cases, there isn't much pleasure to be had by simply moving around (an implicit skill) in quake 3, but the first time you figure out how to rocket jump(an explicit skill) it feels brilliant. The only apparent way to increase the pleasure associated with an implicit skill is to improve the presentation. Examples would be Mario's 'Wa-hoo' when he jumps, or the heady sense of movement you get by simply running in Mirror's edge.

  2. Correction : After reading the project horseshoe writeup again, I got the terminology switched. What I meant that moving around in quake 3 is an 'explicit skill' and rocket jumping is an 'implicit skill'. And the only apparent way to increase the pleasure associated with an 'explicit skill' is to improve the presentation. Sorry about that

  3. Hello,

    I know this is an older post, but I had a question about modeling deep skills using skill chains. I'm currently working on using the concept of skill atoms and chains to improve curriculum design in information literacy, and the field is riddled with deep skills.

    In the appendix of the report from Project Horseshoe you talk a little about deep skills and how some of them never move from mastery to chunking. Rather than be perfect-able, players reach a threshold at which they can use the skill in learning others, but the skill itself is never "mastered".

    I was wondering if you've developed any effective ways of making the distinction between these types of skills and ones that can be perfected in models, or if you've found there to be no practical use for making such distinctions.

    Thanks, love reading your stuff!

  4. Hi Jess,

    Deep skills are a tricky one mostly because they become quite subtle. When an artist makes a distinction between one shade of blue and another, they are often applying some reproducible learning. However, it exists on the level of subconscious instinct and is therefore difficult to either describe or measure as a skill atom.

    You can get quite far in your descriptions (see advanced chess books) but there is a point where the process breaks down.

    Where I've found skill atoms to be most useful are in the skills explicitly supported by the rules and feedback systems of the game. You can often also capture the first level or two of emergent behavior, but as the emergent player skills become more instinct driven and divorced from the existing means of measurement it becomes harder to agree on exactly what the skill atom might be.

    take care,

  5. Hello :)

    Thank you for the insight into where you've found the limitations of the modeling technique to be.

    The way people learn is fascinating, especially when you get into higher level thinking. And you're quite right that the skills as you get farther into learning become quite difficult to define. What's helpful, however, is being able to build up a model of the basic skills that lead to that emergent behavior. Without all the basics in place, it's difficult to move into the high level analysis and evaluation needed for information research.

    Appreciate the help :)