[Wikipedia-l] Showing causation among articles

Mon Jul 12 23:59:55 UTC 2004

Wikipedia has proven itself as an effective process for creating
encyclopedia articles.  But, there's an easy way to give Wikipedia the
ability to to show causal relations among articles.

For example, all articles could have a tab called "influences".  In the
case of the article on Copernicus, users would be free to list hyperlinks
to other articles that complete the sentence

   "Copernicus influenced_____."

and articles that complete the sentence

     "______influenced Copernicus".

If users did this for many articles, then a network of causation would
emerge, where the nodes of the network are articles and the links are
"influences".  This network could be exploited through a search tool that
finds paths of articles between article A and B.

Such a tool would allow for a fascinating study of history.  For instance,
Wikipedia currently has an article on the Cold War and an article on the
Attacks on Sept. 11.  But, what if you want to see if there's a causal
relationship between the two events?  Neither article mentions the other,
and rightly so, since any causal relation between them is too indirect for
the scope of an encyclopedia article.  Also, you won't find one article
titled "How the Cold War Influenced 9/11" because encyclopedias typically
provide information that fills the middle space of the spectrum between
history books on one end, and the daily newspaper on the other.  This
question would usually be a question for history, because normal methods
of tackling such questions require: analysis and time.

But the system I describe above can compress the analysis. Each
influence-link between articles would be one tiny piece of
analysis between events that happpened close together in time.  By drawing
a path through these small links of analysis and time, you can connect
events that happen farther apart in time, thereby automating the analysis.

There are a lot of other uses of this kind of "influence
mapping".  And, you could get a lot of fun data to play with, as
well.  But, that's the main idea I wanted to throw out.  Any thoughts?

Abe