A Closer Look at Wiki Authorship

Jeff Atwood takes an interesting look at the history of changes to wiki pages and the balance between opinion and fact. The larger wikis (e.g. Wikipedia) have a huge amount of data around page edits and Jeff’s article also highlights an IBM study on how the more popular Wikipedia content evolves over time.

There’s also a comment about one of my favourite subjects – reading too much into statistics. Apparently Jimmy Wales (Wikipedia co-founder) looked into who was responsible for most of the articles changes and found that 0.7% of users were responsible for over 50% of all edits. But an “edit” may be a spelling correction rather than adding content or altering the facts or meaning in an article. As it turns out, the data points to these hyper-active users doing just that – cleaning up after everyone else.

Kisimi uses a basic string comparison function called simple_text() to show the relative difference between two versions of a page. We could also use the Levenshtein function which gives the minimum changes to go from string A to sting B, but that doesn’t always make much sense for larger content changes. If someone sees that two versions are 96% the same then it’s obvious they’re much the same.