Let me mansplain this to you slowly...

By George Nichols.

An interesting paper appeared on arXiv (a repository of science and maths preprints) last week studying the significance of various historical figures via the links on biographical Wikipedia pages (http://arxiv.org/pdf/1204.3799v1.pdf ). By ranking biographical pages by the number of pages linking to them and the rank of those pages (in much the same way Google prioritises search results) a table can be drawn up of people's inferred significance and mapped as a 'social network'. The thrust of the paper is the difference in these rankings between the 15 different language Wikipedias and the way these languages are linked. For example, Adolf Hitler tops the ranking in German and French wikis, while George W. Bush takes number one in English with Hitler a lowly third. On a lighter note, Elvis leads the Dutch charts and Pope John Paul II globally outranks Jesus.

All very interesting, but reading the paper one cannot fail to notice a certain something missing from these lists. Of the 25 most significant people on the English Wikipedia there are only two women – Elizabeth II comes in at 23^rd and Hillary Clinton is 25^th. Look at the top five in the 15 languages studied and you will see Elizabeth II ranked 3^rd amongst the Swedes and Marilyn Monroe taking top spot in Norway, but that's it. 75 names and two women.

The authors recognise this startling imbalance and cite the small number of female Wikipedia editors as the cause. This is surprising: women do more volunteer work than men (http://nvs.sagepub.com/content/40/6/1092.full.pdf+html, for example) and the gender gap in internet use has essentially vanished (http://www.pewinternet.org/Trend-Data/Whos-Online.aspx), and yet in 2009 women comprised only 16% of Wikipedia editors, and each was on average half as active as the average male editor (http://grouplens.org/system/files/wp-gender-wikisym2011.pdf). Perhaps more surprising is that this gender gap is not closing over time. As the internet approaches gender parity and its awareness and sensitivity to gender issues has vastly improved Wikipedia remains a huge and hugely influential boy's club.

Lam et al (seriously, go and read that GroupLens paper) find that some articles can be identified as attracting significantly more female (typically people and arts) and male (chiefly science and geography) editors, while other topics are more gender neutral (in that they only reflect the global editor bias and so are still vastly disproportionately male). It turns out that 'female' biased articles are edited more than 'male' ones, and that female editors are significantly more likely to have their early edits reverted - 7% vs 5% for first edits, 6% vs 4% for 2^nd and 3^rd edits, reaching parity between users who have made more than 15 edits (after correcting for vandalism). Since Wikipedia editors whose early edits are reverted are very likely to leave (with no gender bias) women are being driven out of wikipedia by men mansplaining to them that their content is unwelcome.

This situation is really worrying – Wikipedia is rapidly becoming people's first port of call on researching almost any topic. The effect is subtle – few articles actually read as though they are written by a misogynist, but Aragon et al make it clear that when studied as a whole Wikipedia is gender biased to such an extent that only men may rank as important within it's pages, and Lam et al show that female voices are being drowned out. This leads us to the alarming conclusion that the worlds largest and most used reference text and a beacon of free, democratised information contains a remarkably strong and resilient gender bias that is virtually invisible to most users.
Be wary when you Wiki...