OpenStreetMap (OSM) creation in 2004 marked a political, technical and epistemic revolution to democratise the production and visualization of geographic data, pioneering a movement that would later become to be known as Volunteered Geographic Information (VGI)[1]. Many saw in VGI in general, and OSM in particular, as an opportunity to address systemic problems in cartography derived from the historical linage mapping as a means for elites to exert power[2]. Critical and Feminist Geographers have long contended that maps are embodied subjects[3,4] and cultural expressions[5], rather than mere representations. Consequently, maps are never objective or neutral[6], and the historical lack of diversity in map-makers[7] results in biased worldviews when it comes to decide what it is included and silenced[8–10] in cartographic representations, thus reproducing and amplifying further inequalities.
OSM’s signature features (e.g. its global community, open data model and, particularly, its crowdsourced and non-hierarchical approach), provide an ideal sociotechnical infrastructure to overcome those issues. By incorporating a myriad of different worldviews as a result of the ability for anyone to add anything they find relevant to the map, OSM is well-placed to be a cornerstone in the production of equitable maps to allow for other forms of cultural expressions, leading to greater forms of social justice[11].
However, against this techno-optimistic rhetoric (so frequent in the advent of the Web 2.0 in the early 2000s, of which OSM is a paradigmatic example), many scholars and OSM members have flagged that the lack of diversity in the community is still a concern [12–15], despite community efforts to improve the situation. This paper contributes to the corpus of previous literature which has surfaced how factors such as gender, ethnicity or sexual orientation translate into different patterns of contributions. However, previous research has primarily focused only on a particular type of contribution: map edits. While, arguably, this is probably the most attractive and, therefore, the most prominent way of contributing to the project, there are many other ways of doing so (e.g. documenting, participating on community forums, translations or participating in the project’s governance). In this paper, we will be “embracing pluralism”[9(Chapter 5)] by focusing on a different and often overlooked way of contributing to OSM: the tagging proposal process [16], which is the formal process to introduce and discuss how a real-world situation (i.e., a feature) is incorporated (or not) to OSM’s data model[17]. We do so by systematically analysing all the 2,115 tagging proposals that have been documented since 2006 and the 8,487 votes they have received. To allow for a systematic and reproducible analysis, as well as for scrutiny, we created an R package[18] (to be published alongside this paper) that programmatically retrieves all the tagging proposals listed in the OSM Wiki using a combination of API queries and web scrapping. For each of those pages, we retrieved the associated metadata, as well as all the votes received by each of the tagging proposals, and recorded the vote (either “Approve”, “Oppose” or “Abstain”), the detailed rationale (if any), and the user who emitted the vote. The two separate documented datasets[19] produced by this package constitute the primary data source for this paper. From this data, we describe and critically engage with how the tagging proposal process works, the users who are involved in it and what features get proposed, approved, rejected or abandoned. To further engage with existing literature on the impact of diversity in OSM, we expanded these datasets in two ways: first, implementing an automated and reproducible process based on existing literature[20], we classified the proposals into “feminised”, “masculinised” and “other” (i.e., in neither category); second, we followed a semi-automated process to identify users’ metadata about diversity, such as gender, language, and location.
Our initial findings depict a landscape where 998 distinct users (this is, less than 0.01% of the total OSM user base) initiated a tagging proposal – a time-consuming (and mostly individual) endeavour of which only y 27.3% ever reach a decision. On a more positive note, against high abandonment rates, those proposals that do get voted on 73.2% are approved. We contend that because of the challenges of this system, most users only engage with the process of creating a proposal once in their lifetime. However, we identified a reduced subset of 59 users who are extremely prolific and have higher approval rates than other users. Their joint contributions account for 33.8% of the total proposals in the history of OSM, and 44% of the total approved proposals. Furthermore, we find that this group is quite homogenous in terms of gender, geography, and experience: it is heavily dominated by males from Europe and North America with white names and a long history in OSM.
We identified that feminised proposals are significantly less in numbers and systematically receive fewer number of votes than the others (which often leads to not being approved because of not passing the threshold), and when they do receive enough attention, they mainly receive opposing votes.
This research brings a new lens to existing literature on gender gaps in OSM community. The case of tagging proposals is particularly significant because it represents a higher level of participation over map contributions, as these decision-making processes directly shape OSM’s ontology, and thus, the scope of the project. While our findings are in line with prior research, we highlight that problems flagged a decade ago not only are still present but become more accentuated as we climb the participation ladder[21]. Our paper, therefore, puts the focus back on how technology itself -regardless of how well-designed, flexible and well-intended it may be, as is the case of OSM- is not enough to address inequity and power structures. Beyond a description of the problems, we critically engage with the observed phenomenon and, ultimately, propose a series of recommendations in OSM’s governance to make it as inclusive as it aspires to be.