Verfasser  Thema 
Jrep
Nachrichten: 4
Registriert: September 2013

Introducing statistics to Smallworld?

Tue, 29 October 2013 10:47


A few days ago it struck me that the introduction of playing SW online has opened up the possibility of gathering huge amount of data from SW games without too much effort. With enough data it would make it possible to use statistics in order to demystify some aspects of the game that we today know by intuition. Also a lot of fun and creative questions about the game could be answered by statistical analysis.
Im well aware of the implications of starting to look at the game in this way. Some might argue that threatens the core idea of the game. For new SW players this kind of document could possibly ruin the fun of playing the game. However, for experienced players the option to sharpen up your well developed heuristic skills (shortcut appraisals) with statistical data could move you one step further in your understanding of the game.
At this point all of this is just my unfinished thoughts. Ive no idea how much time and work such a project would require. As a psychologist Ive have a decent knowledge in statistical analysis, but I would really need talented statistican who can see the true possibilities of the data. Also, Im not sure what the costbenefit of creating this kind of project would be. As most of you here already have advanced intuitive knowledge of the game it might not be worth the amount of work to get these extra pieces of statistical information.
Anyway let me know what you think, and if somebody gets thrilled by the idea of knowing which score is guaranteeing victory 95% of the time, Which race is most popular by choice or Which race + combos are statisticly most powerful please send me a message for further discussion.
Angry messages about the evil of using statistics in Smallworld are of course also welcome!
Jrep



Lawrence303
Nachrichten: 35
Registriert: December 2012

Re:Introducing statistics to Smallworld?

Tue, 29 October 2013 20:12


I'm on board  I've been collecting statistics from my own games. But how would statistics be gathered? You would need stats from many games to reach valid conclusions. That volume of stats would have to collected by DoW.
A couple of things that have jumped out at me so far (after 68 games):
* I win more games as the second player than as the first player.
* The player who is ahead after two turns loses about 2/3 of the time.



Erik Uitdebroeck
Nachrichten: 1190
Registriert: December 2006



Jrep
Nachrichten: 4
Registriert: September 2013

Re:Introducing statistics to Smallworld?

Wed, 30 October 2013 17:13


Thank you for your input Lawrence, appreciate it!
My idea is trying to get a few experienced players onboard to this project and let them submit stats from their SW games. Another way of doing it is to contact DoW, but I don't think they'll be too keen on giving away data, they might not even have it? However if they would, Any statistical question about SW could be answered in seconds!
Yes, you need to have a big enough sample size to make valid conclusions and how big the size of the sample need to be can be calculated quite easily. Before you can do that you have to decide what kind of questions you want to ask and how these questions affect how we're going to look at the variables.
I like your observations, and I must say that I've noticed anything like that. This is a excellent example that this project must really take a a few players observations into account when creating questions. We see different things when playing! With a sample size of around 100200 games those question would be answered
Please let me know if you think of something more!



Jrep
Nachrichten: 4
Registriert: September 2013

Re:Introducing statistics to Smallworld?

Wed, 30 October 2013 17:18


Collecting data could be the most tedious activity there is on this planet so I completely understand your position there Erik. However my idea is that those who contribute to the project will also share the fruit of it = submitting data!


