Things that interest me.

The “wikilab” at UC Santa Cruz (Go Banana Slugs!) has a demo up for a proposed Wiki reputation system. The system looks at a users edits, and the length of time they survive, and decides that those with long-lasting edits are likely to be more trustworthy. It then allows you to visualize the elements of an article that are written by a trustworthy user, by shading the text more heavily if it is more worthy of your trust.

There are definitely things to like about this approach: it makes use of existing behavior to create the measure, for one thing. But after browsing through some of the articles, it’s clear that the function of the text matters at least as much as the content. If you do some of the wiki-gardening, cleaning up and adding the structural cruft that makes up Wikipedia, you are likely to be judged more trustworthy. Likewise, if you write very broad, general connective sections, you are less likely to be objectionable.

Unfortunately, the most important bits of an article–the ones that really are the most trustworthy–may very well be written back and forth for some time before they reach some level of stability. Likewise, a user may stalk some pretty peripheral parts of Wikipedia and never find herself challenged.

In any case, it is certainly an interesting attempt, worth checking out and thinking about. The demo is here. Just hit “random page” a few times to see some examples of the highlighting.

(via Wikipedia Blog)