Wikidata and Gender

Tuesday 4 July 2017: Wikidata and Gender

In my second training session, last Thursday, I featured a semantic discussion of human “gender”, based on the current view on Wikidata. This would have done little to raise eyebrows of working Wikidatans: a stable position has been reached. The relevant property is called “sex or gender”, and Wikidata controls the acceptable vocabulary to exactly six values.

So far, so good, but surely the whole subject is at the same time vastly interesting and highly contested, as soon as one gets away from sex=gender, and there being a restricted number, two, to go with “he” and “she”? The stated solution takes a minimalist line on one distinction, and more than the fingers of one hand for the other. In short, how come?

I had occasion, during the first training session, to remark of Wikipedia that “the social side is much more complicated than the technology”, something that I wish more people took to heart. Part of that complexity is the way decisions are taken, and consensus positions are adopted. Let me lay it out: (1) no one person made these decisions; (2) any decision, once made, can be reconsidered. These points apply right across Wikimedia communities.

Another point is particular to Wikidata: it all started from scratch. There was certainly no assumption built in that male and female exhausted gender. Properties were introduced one by one, the earliest surviving being P6 for “head of state”. P10 for “relevant video” came along, P14 for “graphic symbol of thoroughfare” started a group of cartographic and geographical properties. P21 is “sex or gender”, rather illogically later than P19 “place of birth”.

Perhaps enough has already been said to illustrate the truth that Wikidata was constructed without a semantic grand plan, whatever the virtues (which are considerable) of the underlying data design. Simply put, everything was up for grabs. P21 went from label “gender” to “sex” and back again on its first day, 4 February 2013, and that was just in English. We can relive all these things, because all the versions are kept. After the version “sex (or gender)” appeared on 26 December 2013‎, the compromise “sex or gender” was seen on 28 January 2014‎. That stuck.

Of the four non-traditional permitted values, genderqueer (Q48270) is logically the third, meaning “neither of the above” when below male and female genders. So, less exciting than it may sound. Then intersex (Q1097630) relates to non-standard sex characteristics, while transgender male (Q2449503) and transgender female (Q1052281) relate to self-identification. Two of these being catch-alls, there is a certain solidity to the solution adopted. But never say never also applies, to the question whether a consensus will change it in time.

I had a comment after the talk part of the session, in which I’d gone over this ground quickly, to the effect that the discussion page backing up P21 was on the reasonable side. That is good to hear, at least if it means enough constructive discussion went on to document a debate on options that touched enough of the bases.

What does this all prove? It certainly doesn’t ensure that some edge case or new perspective won’t at a future point upset the applecart of the current consensus. It doesn’t prove that the “crowd” – the apparently random aggregation of Wikidata editors who contributed – came up with a de facto standard, which should then be adopted more widely. A decentralised discussion doesn’t often do that.

Some unkind things are said of design by committee, and what results from Wikimedia social processes is in the same vein, though asynchronous. In other words, things get do hashed out, over time. Perhaps a saving virtue is that having the moral courage to disagree with consensus on Wikipedia is much less scary than in many work or social situations. On the other hand, divisive rhetoric and politics thrive on re-opening questions, deploring stability as much as compromise. It can be heartening simply to see their sway isn’t universal.

Here is the links page for the session on 29 June 2017.