What happens when VGI is threatened? A systems perspective analysis of the events behind the introduction of rate limiting in OpenStreetMap (sotm2024)

This talk presents a systems perspective analysis of events in October-November 2023 which started with large-scale vandalism of OpenStreetMap data in Israel and ended with the introduction of rate limiting. Noting how the project reacted in face of this unique situation and the undermining of its basic assumptions facilitating the project helps uncover resilience- and vulnerability-inducing mechanisms within it, to characterize the range of possible external influences on the project, and assess their possible outcomes. OpenStreetMap (OSM) is generally considered to be a ‘do-ocracy’ [1], a governing system in which power stems from doing and those who do more (contribute more) have greater ability to influence the project. This does not mean that hierarchy within OSM is entirely bottom-up, as Fast & Rimmer’s systems perspective model of a Volunteered Geographic Information project suggests [2] - it is made of interconnected components (the technical infrastructure, project, and contributors) each providing a different level of direct access to the core of the project. Entities may move between these components, but a certain level of separation still exists, enforced informally or formally by gatekeeping procedures or individual acts. The fuzziness of borders and the interdependence of components open for multiple types of interactions that may translate to major impacts on the nature of the project. In this talk we use communication records from various media (the OSM_Israel Telegram group [3], community forum discussions [4,5], and GitHub issues [6,7]) to: i) to analyse a specific set of events - the politically-motivated edits in Israel and the subsequent introduction of rate limiting in OSM [8] - from a systems perspective; and ii) conceptually explore power dynamics and impacts within collaborative geodata projects and specifically OSM along with their implications for the project’s vulnerability and resilience. The chain of events started on 20 October 2023, when two members of the OSM_Israel Telegram group [3] reported receiving offensive comments on their recent edits. One of them noted that the commenting user was performing edits to the OSM data in Israel that were clearly acts of vandalism, i.e. deleting and distorting data. It was only the day after that the members of the Telegram group realised that these acts were part of a greater effort in which 3 newly registered OSM users were performing large-scale deletions and distortions of data in Israel. The timing of these edits, taking place while the Israel Defense Forces were launching airstrikes over Gaza strip in retaliation of the 7 October attack by Hamas, and their related changeset comments, e.g. “There is no country Israel & Free Palestine”, clearly attested that they were politically-motivated (For a representative example, see [9]). The events were quickly reported to the OSM Foundation’s (OSMF) Data Working Group (DWG), the voluntary body responsible for handling data issues such as copyright infringement and vandalism. The DWG issued bans to involved users, yet these were replaced in the next few days by more new users together making tens of thousands of similar edits. Meanwhile, one user, responding to these acts, further vandaled the map by ‘annexing’ part of the Gaza strip [10]. In the immediate aftermath of these events, most of the OSM data for Tel Aviv was deleted from the map and data all across Israel and the eastern Mediterranean basin, as far as Crete, Cyprus, and Turkey, suffered from distortions (the scripts the vandal accounts were using automatically distorted random objects crossing the area of Israel, including objects such as coastlines that extend beyond these boundaries). Yet the impacts of the events did not end with the DWG reverting the vandal edits. These events had a direct connection with the introduction of a daily rate limit (set individually based on the seniority of the editing user and the number of edited entities, among other things [8]). The idea of such a limit had already existed for a while and was discussed within dedicated GitHub issues [6,7], yet it was only after members of the Israeli OSM community had commented within these issues and in a related community forum discussion [5] that the notion had been implemented. Interestingly, a DWG member which was handling the bans and reverts was the one who referenced the Israeli community to the forum discussion - in response to a question raised in the Israeli Telegram about whether any measures have been implemented to ensure this would not happen again, this user had suggested that they will ask that question in the relevant forum discussion. As extreme as this vandalism case was in comparison to other acts of vandalism or politically-motivated edits, its outcome tells a story regarding the fluidity of power dynamics within the project and accordingly - its vulnerability- and resilience-inducing mechanisms. The enforcement of rate limiting required a temporary coalition between individuals within the contributors and project components, where DWG members used the ‘traditional’ crossing path - from project to contributors, i.e. the DWG engaging with contributions to the point of joining the Telegram group - to mobilise support for their cause by encouraging contributors to engage with the management of the project, i.e. joining the GitHub discussion on rate limiting. This collaboration was in turn enabled by a joint sense of urgency stemming from the vandals violating the basic assumptions required for OSM to function - shared goals, adherence to community procedures, and mapping limited to verifiable objects. On the one hand, the temporary coalition formed between the local OSM community and the DWG members shows that the fuzzy nature of component boundaries allows for flexibility that enhances resilience in the sense of the ability to reorganise and adapt when conditions change. However, it shows that even temporary coalitions (it is useful to remember the Israeli community’s dissatisfaction with how the DWG handled the Jerusalem node issue [11]) of a handful of people reacting to a local event can lead to a global change occurring without a general consensus within the community (some dissatisfaction with the decision was already expressed [12,13]). If this becomes a common practice, it may undermine the stability of community procedures within OSM, especially if disagreements emerge due to clashing interests. Beyond this specific insight, this case allows mapping out different axes of power dynamics within OSM - the stakeholders (i.e. the system components), the crossing paths (e.g. project-to-contributors, contributors-to-infrastructure), the sources of capital (e.g. the masses, the project, access to resources and/or infrastructure), the nature of crossing (e.g. invitation, infiltration, takeover, acquisition), and motivations (e.g. individual/communal interests, safeguarding the project, values). Characterising other cases of external or semi-external engagement with OSM in light of these, e.g. corporate engagement with OSM, the collaboration of Meta and HOT (The Humanitarian OSM Team), and the development of Rapid, can produce insights on their deeper implications for the project and help imagine future paths of influence and their impacts. Creative Commons Attribution 3.0 Unported https://creativecommons.org/licenses/by/3.0/ about this event: https://2024.stateofthemap.org/sessions/FSYR9S/