[Wikipedia-l] Stable versions policy

Thu Dec 22 14:29:01 UTC 2005

Wikipedia Romania (Ronline) wrote:

> The fuss is about the fact that under the stable versions 
> policy, Wikipedia will change from becoming a fully open, free 
> wiki encyclopedia, to a static encyclopedia with a wiki 
> *option*. It reminds me of both the failed Nupedia and the 
> structure Encarta announced a few months ago to make itself more 
> "open".

I might not agree with your analogies, but if you voice this 
criticism now, others are likely to do the same, and that means it 
would be wise to have the counterarguments ready.  Instead of 
saying "we should have stable versions", we should state exactly 
what the current problem is, so we can tell if the new approach 
succeeds or fails to solve it.

In the beginning, someone must have thought "if I only had a free 
online encyclopedia...".  What we have now with Wikipedia might 
not (yet) fulfill the need of an (all-round) encyclopedia, but it 
certainly is a database of many good articles, from where you can 
often pick what you need.

Instead of thinking of (the English) Wikipedia as a basket of 
850,000 articles in their current version, we can think of it as a 
basket of X million article-versions.  Maybe I like the latest 
version of the article on Gabon, but the December 3 version of the 
article on Angola.  For another article, say Vadsbo (a stub), I 
might be frustrated with every version, and this is no different 
from a topic where Wikipedia has no article at all.  Every 
article-version in this collection of X million is free under 
GFDL, and can be used as the basis for new article-versions.  
Anybody can add new article-versions to the collection.

Currently Wikipedia works under the assumption that the latest 
version of any article is the best one, and the only one that 
should be linked from other articles, shown to the public and 
indexed by search engines.  It is important that we realize that 
this *is* an assumption, it does represent a design choice, and 
not necessarily the optimal one.

If I'm compiling a Wikireader or a similar subset of articles, I 
might pick my article-versions under very different assumptions.

I'm surprised that Wikipedia mirrors such as Answers.com don't 
work more like Wikireaders, where a human editor picks useful 
article-versions and leaves the stubs unmirrored.  The added value 
from such an "editor's choice" would be a perfectly valid business 
model.

-- 
  Lars Aronsson (lars at aronsson.se)
  Aronsson Datateknik - http://aronsson.se