webdiplomacy.net

Is there a database somewhere of individual games and the detailed data from the games (moves, results, etc). that is available publicly for analysis and the like?

mhsmith0 wrote: ↑

Fri Jan 05, 2018 9:18 pm

Is there a database somewhere of individual games and the detailed data from the games (moves, results, etc). that is available publicly for analysis and the like?

God, I hope not. No good would come of it.

I tried for a while to get some data from the admin team to do some data science shit but it was a painful experience and I gave up.

You're the same guy who said "I'd contribute to the code, but you wouldn't accept the pull request", right?

Anyway, @mhsmith0 - not exactly, but also yes. It's possible to get a database dump of just the public information, but it's kind of annoying to generate (someone has to write a query that gives you a valid database without exposing private information like press or other user data).

If you're super interested and prepared to do a bit of leg work, you can see the database structure in the github repo - if you write a query that will pick up what you want without exposing private data. I'd be happy to run it for you. Unfortunately I don't have the time to put together the queries, otherwise.

Other options if you don't want to do that are to ask the current admin team (I don't work here, I just have database access in case anything catches fire), or to construct it based on the data you can see as a logged in user (but please be polite in any screen scraping you do).

I'm always interested in the results of anything done with the webdip data. I did some analysis a while back which is buried deep in the old forum somewhere.. I also did some generations of solo maps like these that I thought might be interesting, but they pretty much showed the same picture.

Oh, also, I made a docker harness that you might find useful if you want to see the database schema without running all those scripts yourself - you can find it here.

I haven't tried this, but you can probably do a docker run of to get straight into a DB console.

ATC if you would run queries I'd be delighted. I tried to get some form of database access from HR but it went nowhere.

ATC, it is unlikely you will find much difference in the stats from here to whereever Josh Burton game stats came from. At some point, the sample size is going to even everything out. I sincerely doubt the platform would make much difference.

What would be interesting would be to find data only on games with experienced players involved. It would probably not be much different, but I am curious to see if there is any significant difference in outcomes.

I actually suspect there would be substantial differences, particularly in gunboat. Austria becomes viable instead of toast, Turkey becomes harder to succeed at, etc.

Condescension wrote: ↑

Wed Jan 10, 2018 2:27 am

I tried to get some form of database access from HR but it went nowhere.

I went ahead and underlined the main source of all your problems.

bo_sox48 wrote: ↑

Fri Jan 12, 2018 5:59 pm

Condescension wrote: ↑

Wed Jan 10, 2018 2:27 am

I tried to get some form of database access from HR but it went nowhere.

I went ahead and underlined the main source of all your problems.

Yeah, I didn't really know who's responsible for what. I could've done without all the leg pulling though.

It wouldn't be appropriate to give out database access - it's not designed to be given out, and the site has a reasonable responsibility to keep players private data well, private. Plus the only way to access the database is via shell access to the account that hosts the site.

The "only experienced players" thing is interesting, but I'm not sure how best to measure experienced players. GR is certainly not very predictive (I wrote a longer post on this on the old forum).

I do think that webdip play comes in fashions, especially among players who are very active. There was a time where france was doing well (and Italy poorly), because few italian players were going to Piedmont. I haven't hit on a good way to analyse that.

By far the most interesting analysis was looking for "natural allies" - countries that tend to both do well in pairs. The correlations were not strong, and then the results were not particularly surprising - for example, France and Austria tend to do well together (probably in games where italy has done poorly). Even then, the correlations weren't strong enough to make big conclusions about.