User talk:Knulclunk/Random

How random is Special:Random?

I'm not sure Special:Random gives a representative sample.

One day I clicked it a lot and it had a huge percentage of Association football-related articles. Another day I clicked a lot and there were many fewer such articles. It would be nice to know just how random "random" really is. davidwr/(talk)/(contribs)/(e-mail) 03:30, 2 February 2009 (UTC)[reply]

Man, talking about statics, what is the chances of getting an edit conflict on a red linked talk page? I just did. LOL Ikip (talk) 03:32, 2 February 2009 (UTC)[reply]

High, if the page is PUMPED.

Copied from other talk pages

From: Wikipedia talk:WikiProject Countering systemic bias/Archive 12

From my own curiosity, I calculated article topic distribution for 200 random articles. The results are interesting and can be found here.--Knulclunk (talk) 14:27, 24 September 2008 (UTC)[reply]

In all honestly, anyone with minimal stats knowledge would probably talk about how 200 is far from a good sample. But it's still interesting...though the results are hardly unexpected to me anyway. ♫ Melodia Chaconne ♫ (talk) 14:51, 24 September 2008 (UTC)[reply]

You have to distinguish cases. The pie chart shows 36% of articles about the US. That means 72 out of 200. By simple rule of thumb, the margin of error is roughly ± square root of 72(This computer doesn't seem to do the symbol), or ±8½. That's enough to make clear the excessive emphasis on US. For smaller nos. your remark is correct. Peter jackson (talk) 15:35, 24 September 2008 (UTC)[reply]

Oh those statisticians... alway ruining a good party. I figured I would keep going to see if distribution leveled out at 500 or 1,000. I wonder what percentage of 2.5 million do you need to be within ±3.5%? --Knulclunk (talk) 15:41, 24 September 2008 (UTC)[reply]

What about articles on clearly international topics (such as lunch or school) or non-Earth topics (such as planets)? Could you also take a random sample of FAs? --J.L.W.S. The Special One (talk) 04:45, 1 October 2008 (UTC)[reply]

What a great idea. I was thinking earlier that this kind of statistics should be done more often, maybe on a yearly basis to assess the evolution of Wikipedia articles. We could make a bot or a program which will generate a list of 1000 random articles, and I could create a program which will make their analysis and their count simpler. Eklipse (talk) 08:41, 1 October 2008 (UTC)[reply]

Eklipse, I think that your proposal to semi-automate this process is very worthwhile. --Wormcast (talk) 20:47, 12 December 2008 (UTC)[reply]

Yes, excellent job! I actually just suggested that exactly this type of survey be done, only to find out that you had already done it. My only gripe is that I can't discern some of the colors on the 'Type' chart; e.g. Geo. Political and Geo. Landforms. --Wormcast (talk) 20:41, 12 December 2008 (UTC)[reply]

Ikip (talk) 03:32, 2 February 2009 (UTC)[reply]

Interesting

Hey there Knulclunk. Just wanted to let you know that I found this experiment interesting and I've decided to add it to the Random Pages Test category so it can recieve wider exposure. I hope you don't mind. -- œ ^™ 22:45, 17 September 2009 (UTC)[reply]

Cool! Thanks. :) --Knulclunk (talk) 22:54, 17 September 2009 (UTC)[reply]

I've noticed a pattern too with my use of the random article feature. I keep getting articles about villages in Poland and Iran (occasionally Canada and Romania). – Illegitimate Barrister (talk • contribs), 03:43, 11 February 2018 (UTC)[reply]