Zooming In on Agreement-Making: Tracing the BBNJ Negotiations with the MARIPOLDATAbase

On June 19, 2023, more than 100 governments adopted by consensus a new international legally binding instrument under the United Nations Convention on the Law of the Sea (UNCLOS) on the conservation and sustainable use of marine biological diversity of areas beyond national jurisdiction (BBNJ). After five intensive years of negotiations marked by several conflicts over BBNJ regulation, governments agreed on a “package deal” including measures regulating marine genetic resources, including questions on the fair and equitable sharing of benefits (MGRs), area-based management tools (ABMTs) including marine protected areas (MPAs), environmental impact assessments (EIA), and capacity building and transfer of marine technology (CBTMT). From the beginning of the negotiations, research interest in the BBNJ process was high (de Santo et al. 2019; Tessnow-von Wysocki and Vadrot 2020; Vadrot 2020) and it is continuing to grow in anticipation of the agreement’s entry into force, which is anticipated during the 2025 UN Ocean Conference (IISD 2024). This research note aims to assist future BBNJ research by introducing the MARIPOLDATAbase, which systematically covers all rounds of negotiation (2018–2023). Global environmental politics research has strongly benefited from studies using data on multilateral negotiations collected on-site (Campbell et al. 2014; Vadrot 2020). “Being there” (O’Neill and Haas 2019) leads to insights into the processes, actors, and sites of global environmental agreement-making (Hughes and Vadrot 2023; Hughes et al. 2021) and enables solo researchers and research teams alike to unpack the complexity of global environmental meetings from various research perspectives (Chasek 2001, 2019; Gray et al. 2023). Dealing with processes complements what researchers focusing on environmental negotiation outcomes—“agreements, protocols, or amendments that states successfully negotiated”—can do: comprehensively review the development of international environmental agreements and how they have changed over time (Mitchell et al. 2020).

Yet, instead of focusing on the broader architecture of international environmental law (Mitchell et al. 2020) and its meta-structures (Kim and Morin 2021), the MARIPOLDATAbase addresses the micropolitics of negotiating a new agreement; this is achieved by mapping and tracing the entire stream of negotiations that took place in negotiation rooms—from the first intergovernmental conference in 2018 to the closing session in June 2023. Meetings behind closed doors, which increased in later stages of the negotiations, could not be observed and are not covered in the MARPOLDATAbase. Our seven researchers observed six intergovernmental conferences (IGCs) and forty-five intersessional meetings in all, both online and on-site, and the two-day closing IGC. Our field note–taking method was built around a collectively designed Excel file recording our observations (Vadrot et al. forthcoming).

We approach international negotiations with a collaborative event ethnographic (CEE) perspective, focusing on the processes rather than the outcomes of international negotiations in which governmental and nongovernmental actors shape the negotiation outcome. However, it is different from other CEE contributions in that we use participant observation and collaborative field note taking as a data collection method and analyze the resulting data with a number of different methodological approaches (Vadrot et al. forthcoming). This separation of data collection and analysis methods has the advantage of allowing different researchers to contribute to the same database of observations that is then used for different research purposes. Thus, while we applied this approach to serve our collective and individual research interests in the negotiation of the BBNJ agreement, we also attempted to organize data collection and management in a way that would allow us to make our data available to those who were unable to be on-site and closely follow proceedings (Vadrot et al. forthcoming). In addition, we wished to increase the value of ethnographic data collection and support the application of mixed-method approaches by providing data that could be used for several analytical purposes and various sets of research questions.

The MARIPOLDATAbase traces the BBNJ process across different meeting formats, including plenary sessions, working group meetings, “informal informals,” side events, and the online intersessional period, which, owing to the COVID-19 pandemic, extended over two years (Langlet et al. 2022; Vadrot et al. 2021; Vadrot and Ruiz-Rodríguez 2022). The database covers different meeting formats where observers were allowed to participate, including in an anonymized manner in formats such as informal informals, which had a formal procedure but were confidential and accessible only by registered participants, and online dialogues. But the MARIPOLDATAbase does not cover fully closed meetings, such as “president’s consultations” or any bilateral meetings among negotiating states or state groups. Thus the MARIPOLDATAbase also reflects the modalities of negotiating, that is, where observers were let in the room and where they were not, and describes the BBNJ process from the viewpoint of a nongovernmental organization (NGO) observer. It consists of more than 43,000 entries and allows the use of primary data for BBNJ- or negotiations-related research. We hope that the MARIPOLDATAbase will become a major resource for scholars, students, and practitioners around the world seeking to understand BBNJ agreement-making, actor constellations, networks, state-specific information, practices, discourses or narratives, or trends in current global environmental negotiations.

In what follows, we introduce the MARIPOLDATAbase and methodology, then describe how data were collected and the database created. We will navigate readers through the database, illustrate its contents, and exemplify analyses that can be conducted with the data. Finally, we will identify potential research questions that could be addressed by using the data and reflect on how our CEE approach has evolved.

The MARIPOLDATAbase is a central output of a six-year research project, MARIPOLDATA (2018–2024), financed by the European Research Council (ERC), involving a team of seven mostly early-career researchers led by Alice Vadrot. One of the objectives of MARIPOLDATA was to trace the BBNJ negotiations by collecting data through CEE from the very first meeting until the adoption of the final text. Originally, the mandate for these negotiations was to reach agreement in the course of four IGCs. Owing to the COVID-19 pandemic and complex, slowly moving negotiations (Mendenhall et al. 2019), however, an additional IGC 5 was halted in August 2022 and continued in February–March 2023 as IGC 5.2, where states finally reached agreement. The final adoption of the agreement took place in a ceremonial resumed IGC 5.3 (adoption conference) in June 2023. As a result, the overall participant observation and data collection phase had to be extended until June 2023. By shedding light on the making of the MARIPOLDATAbase, we will show how we addressed two key challenges encountered during CEE: taking field notes during negotiations through participant observation as a team and cleaning and managing our data for reuse by researchers outside our research group.

Large-scale global environmental meetings challenge the practice of participant observation and data collection: what should be observed or recorded, how should field notes be taken, and how can the consistency of work by research team members be ensured? CEE scholarship has addressed these issues by recommending that a table be introduced and populated to allow the entire research group to develop a shared understanding of what to observe (Campbell et al. 2014, 11) and how to record it. While this approach may facilitate research by, and communication within, a research group, it does not allow “outsiders” to benefit from on-site observations, a challenge that is generally encountered as regards (ethnographic) data on multilateral environmental negotiations collected during the event(s). Reconstructing negotiation dynamics for the benefit of a researcher who has not attended a meeting is almost impossible; it often requires working with secondary data and mixed-methods approaches (Hughes and Vadrot 2023). Here the Earth Negotiations Bulletin (ENB) reports have proven to be of extraordinary value to the research community by providing daily summaries, including pictures and impressions of what happens “in the corridors” (Allan and Chasek 2023). Furthermore, a number of studies have been published on the BBNJ negotiations, including a special series of articles on each IGC in Marine Policy (Mendenhall et al. 2019, 2023; Santo et al. 2020; Tiller et al. 2019). But neither outlet counts as a systematic record or primary data that might be used to empirically study the negotiation process.

To address the challenge, we developed an MS Excel file that would streamline what was observed and recorded and could be used by all team members. In line with Emerson et al. (2011), we focused on the “data” of fieldwork by making sharp distinctions between recording what “others” (state actors, non–state actors, secretariat staff, and other people in the negotiation room) said and did and specific notes incorporating our thoughts and reactions to what we observed and how we interpreted it (see Vadrot et al. forthcoming). For all sessions (on-site, online, or hybrid), our team used the same field note–taking template and method: a matrix—an Excel sheet with predefined categories—streamlining what is observed and how events, observations, and conversations are recorded (Vadrot et al. forthcoming).

In the MARIPOLDATAbase, one observation is not necessarily one statement (or speech act), as note takers sometimes extended a statement across a number of rows to mark changes of the issue addressed. Thus, while there are 43,341 rows of observations, these constitute only approximately 23,635 individual statements. The running_id has an id per each observation/row and the id_combined marks changes in the speech act, that is, one id_combined is always the same statement and sometimes stretches across a number of rows. Note takers were free to record information across new rows belonging to the same statement and generally did so when the speaker addressed different provisions or issues.

Most of the observations were taken during the 5 + 2 IGCs, but many observations were also made during the online dialogues and intersessional work (for a review of the online negotiations that took place during the COVID-19 pandemic, see Langlet et al. 2022; Vadrot and Ruiz-Rodríguez 2022).¹ According to the modalities of negotiations, during the first three IGCs, most observations were taken during plenary or working group sessions. When the modalities changed, and from IGC 3 onward, negotiations increasingly moved to informal settings, and a growing number of observations were also taken during informal informals. Side events were observed only during IGCs 2 and 3 because the research team did not directly attend IGC 1 and, from IGC 4 onward, no side events were organized (see Figure 1).

All observations taken during informal informals, informal dialogues, and online dialogues/webinars/briefings (approximately half of all observations in the database) were systematically anonymized so that database users can access the content of what was said but cannot view any information on the actor or group that made the statement, according to the Chatham House Rule.² The anonymization was conducted in a systematic way that allows analyzing the negotiations without knowing the actor making a statement. Each actor was assigned a random number (i.e., “a state name” became anonymized_state_68, “another state name” became anonymized_state_181, “an IGO name” became anonymized_IGO_10, etc.), which remained the same throughout the database (i.e., anonymized_state_68 remains anonymized_state_68, anonymized_IGO_10 remains anonymized_IGO_10, etc.). In this way, a user can reliably and systematically trace positions and statements while the actors remain nameless. This anonymization method was chosen to balance the wish for in-depth research with the requirement to anonymize observations from informal negotiation settings, which is further discussed later. At the same time, all observations from open negotiation settings, such as working groups, plenaries, and side events, contain all information, including on the speaking actor. A fully nonanonymized anonymized version of the database can be shared for research purposes upon reasonable request.

Overall, we collected data on 108 separate days, the first observation being on 2018-09-04 (first day of the first IGC) and the last on 2023-06-20 (the final day, when the IGC adopted its report and concluded its work). Figure 2 depicts the growth of the database, starting with IGC 1, followed by a COVID-19-induced break, after which the research team observed intersessional work and online dialogues.

By the end of the data collection, the team had filled 201 Excel files, grouped by IGC or by intersessional dialogue. To clean and structure this large number of raw data files, we developed a code in the programming language R that merged all data, sorted it in chronological order, and performed several cleaning and data manipulation steps.³ These included detecting and correcting many typing errors (made during data collection) in several columns and a ChatGPT4-supported spell-check of the descriptions, which was added as an additional, “clean” column next to the originally typed text. These data wrangling and cleaning steps allow for effective filtering and browsing of the data.

The MARIPOLDATAbase contains 42 columns (variables; see Table 1). To allow intuitive browsing of the database, they are organized in the following way: first, a description of the observation; second, the content and its positioning within the negotiations and the BBNJ draft; third, contextual information that allows tracing back and reconstructing the proceedings; fourth, analytical variables that the research team prepared; and, finally, metadata. Documentation is accessible with the database describing the content and method of collecting and cleaning each variable in detail.⁴

The structure of the database allows users to easily filter every relevant statement or to display an aggregation of observations, for example the number of observations per package item (see Table 2). One of the strengths of the database is that the provided data make it possible to combine several research approaches to study the object of interest holistically. This filtering ability could also be adapted to support additional research interests. For example, as a research team, we jointly focused on research questions related to science. For this reason, a column was added to the MARIPOLDATAbase where a box could be ticked every time an actor made a reference to science, including directly referring to the terms science/scientific or mentioning and explaining scientific concepts/terms, referencing scientific institutions or scientific outputs/events, and naming scientific research activities and capacities (Tessnow-von Wysocki and Vadrot 2024). In this way, the MARIPOLDATAbase enabled filtering for all individual statements across IGCs on one specific topic.

The MARIPOLDATAbase can be used to cut through the complexity (Chasek 2001) of BBNJ negotiations: the numerous actors, alliances⁵ formed, positions taken, statements issued, networks created, concepts referred to by actors and included in the text, and meeting formats used. So far, we have used our data, inter alia, to analyze the struggle over the common heritage of humankind principle (Vadrot et al. 2022), the emerging BBNJ regime complex (Langlet and Vadrot 2023, 2024a), and the transformative power of the notion of “ecological connectivity” (Tessnow-von Wysocki and Vadrot 2022). The following section gives an overview of potential research avenues and themes that could be pursued using the MARIPOLDATAbase.

As with other multilateral negotiations, non–state actors, most notably NGOs (e.g., World Wildlife Fund [WWF], Greenpeace, or the High Seas Alliance—a network of NGOs), scientific institutions (e.g., International Union for the Conservation of Nature, Deep-Ocean Stewardship Initiative), and industry bodies (e.g., International Cable Protection Committee) played an important role. While statements issued by states during the IGCs do account for the greatest percentage, non–state actors took advantage of the space given to them during the intersessional period (see Figure 4), where experts were invited to talk about specific aspects of the agreement (e.g., financial mechanisms or MGR traceability). Figure 3 shows that the presidency and the conference secretariat issued an increasing share of the statements.

If we look at activity from the individual state, as in Figure 4, we gain a more detailed picture. Most interventions were made by the European Union (EU), the United States, and Russia, followed by China. This needs to be read with some care. Many states participated in the negotiations as part of an alliance, whereby one state would speak for the whole alliance or one state would represent the whole alliance for each package item. The following graph merely indicates that participation in the BBNJ negotiations was an (almost) global endeavor.

Many states aligned themselves with a regional group. To capture this important aspect of multilateral negotiations, we entered this information into our Excel file by including two variables: alliances, to indicate membership in an alliance by a state actor, and obo—each time a state actor spoke “on behalf of” (obo) a particular alliance, we recorded this type of behavior.

To capture the temporal dimension of negotiations, we recorded the start and end times of each observation; this indicates how long specific actors spoke for and how dynamic the interaction between actors was. Our data indicate large differences between meeting formats and negotiation stages. During the opening of plenary sessions, we recorded very long statements with almost no interaction; during informal informals, we could observe very short and lively forms of interaction between actors.

Figure 5 illustrates the fact that although the EU made the most interventions as an individual actor, other alliances spoke more in terms of time. The United States and Russia were the two individual states that spoke the longest. One can also see the varying emphasis on package elements of the agreement. While the EU spoke mostly on EIAs, China invested a lot of time taking the floor to discuss MPAs, and the African Group spoke mostly on MGRs.

To capture the content of statements and how they relate to the agreement’s text, its package elements, and its provisions, we recorded statements as such (by transcribing as much as possible)⁶ and entered information into the Excel file about the part of the text under discussion. Thus the MARIPOLDATAbase includes detailed information about the contents of statements and their relationships to the various versions of the BBNJ text. It assigns each observation to a package item, the title of the section and article of the draft being negotiated at the time, and the number of the subsection and issue (issue_detail) under negotiation.

The database also contains a link to the draft text (draft) that was being negotiated at the time; this allows researchers to look into the draft at that point in time and find the content of the provision under discussion.

In addition to detailed information on the specific draft provision under discussion, the MARIPOLDATAbase contains information about the option (option detail) that was being discussed, as well as the position of a given state concerning this option. To this end, we included the variable pro_contra, which indicates whether a state was for, against, or neutral to a provision.

Such a listing of state positions on specific provisions can facilitate analysis and comparison throughout the entire negotiation process. By zooming in on specific issues, these data can support the analysis of situations of dissent by providing an overview of who was in favor of or against a specific aspect of the agreement text. We relied on these data when we analyzed the struggle over the common heritage of humankind principle (Figure 6), where we combined it with a qualitative analysis of the statements themselves (Vadrot et al. 2022; see Figure 7). This example demonstrates that the MARIPOLDATAbase is well suited for combining qualitative and quantitative analysis, for it not only maps and traces the course of the five-year negotiation process but can also zoom in on specific moments.

The MARIPOLDATAbase is suited for social network analysis. Our understanding of “negotiation sites as organised social spaces, in which regime complexes form through interaction and recognition between state and international organizations (IO) actors” (Langlet and Vadrot 2024b), is reflected in the MARIPOLDATAbase, which allows researchers to zoom in on relations between actors. States not only mentioned each other but also commonly referred to IOs—network analysis can reveal patterns in these references. A relational analysis can use such references to create network maps.

Relational data can also be derived less directly. Whenever states (or IOs) make statements on the same provision, a link between these two actors can be created: it expresses that both actors have a position on or interest in the same provision. In this way, researchers can analyze the overlapping interests of different IGOs in the same provisions (Langlet and Vadrot 2024b, see Figure 8). While these analyses can technically also be conducted with the anonymized parts of the database (all observations from informal or online negotiation settings), they are more meaningful when researchers can attribute the actor to a statement; thus the following examples use data that were not anonymized from plenary and working group negotiation settings in IGCs 1–3.

Overall, a large amount of information from the MARIPOLDATAbase can be used to produce network maps, which can be interpreted with the help of the qualitative depth of the data (Langlet and Vadrot 2024b). This approach to using data is particularly valuable because network analysis tends to neglect much information contained in each observation by taking out the relational aspects from a large number of observations. However, the data make it possible to complement and deepen the network graphs more qualitatively.

A qualitative approach to analyzing data may require a closer look into statements. Thanks to its structure, the MARIPOLDATAbase enables quick filtering and search of the relevant sections.

Generally, opening and closing statements are longer statements and may be of special interest for qualitative discourse analysis because they consist of longer declarations during which states eloquently display their lines of argumentation and express their views. They can easily be selected by filtering the type of observation (type_obs = = 3; see Figure 9).

Language can also be explored quantitatively, for example, through so-called sentiment analysis. The database contains two variables that calculate the sentiment for each statement on a positive–negative scale (sentiment_afinn) and an unconstructive–constructive language scale (sentiment_constr); this was built using a latent semantic scaling model (Watanabe 2021). Sentiment analysis was used to calculate and compare the “constructiveness” of statements made at two IGCs in a MARIPOLDATA blog post (MARIPOLDATA 2023), but it could also serve to analyze the “constructiveness” of language throughout all IGCs (see Figure 10).

Figure 10 shows that states engaged in ever more constructive discussions, starting with IGC 3, which seems to correlate with the arrival of informal informals.

Monitoring the use of ideas in negotiation settings can yield insights into the framing of specific issues, interests, meanings, and values underpinning agreement-making, for instance, when states discuss the inclusion of concepts into legal text (e.g., “ecosystem services,” “ecological connectivity,” or “biocultural diversity”) (Gray et al. 2014; Hughes and Vadrot 2019; Kobayashi et al. 2020; Vadrot 2020) or when a concept with a specific meaning in one international negotiation forum is introduced in another (e.g., “common heritage of humankind”). The MARIPOLDATAbase can be used to follow the paths of ideas or topics through the BBNJ process because it allows users to search for terms and aggregate data accordingly. This research avenue has already been embarked upon using this data set to gain insights into the notions of “ecological connectivity” (Tessnow-von Wysocki and Vadrot 2022) (see Table 3) and “common heritage of humankind” (Vadrot et al. 2022), as well as the topics and ideas that traveled into and across the digital multilateral sites during the intersessional period (Vadrot and Ruiz-Rodríguez 2022).

As we recorded our observations on how Global South state actors struggled to change the meaning of the common heritage of humankind principle, users of the MARIPOLDATAbase can also follow this process and contribute to its research with new insights. UNCLOS indicates that all states shall benefit equally from economic benefits derived from the exploitation of seabed minerals (nonliving resources) in areas beyond national jurisdiction (Vadrot et al. 2022). Yet Global South state actors at the BBNJ negotiations pursued including within this principle the sharing of the benefits arising from the exploitation of MGRs (living resources) obtained from the water column and seabed in areas beyond national jurisdiction. Owing to the fact that Global North and Global South state actors did not reach agreement on this new meaning, the BBNJ agreement upholds the original meaning of the common heritage of humankind principle as indicated in UNCLOS.

With the onset of COVID-19, IGCs for the BBNJ agreement were postponed. NGOs quickly adapted to the new digital meeting environment and initiated online dialogues to continue discussions on elements of the draft, while the BBNJ Secretariat offered digital tools to facilitate communication between states (Vadrot et al. 2021; Vadrot and Ruiz-Rodríguez 2022). The research team adapted to this unprecedented situation and continued taking notes, this time during online meetings (intersessional work and online dialogues).

Thus the database contains a considerable number of observations (∼7,500) on digital dialogues and digital intersessional work. This mix of observations, taken both at on-site and online meetings (indicated by the negotiation_format variable), makes the database a particularly valuable resource for approaching negotiations taking place online, through digital communication tools, which are often difficult to access, and for comparing on-site with digital negotiations across several indicators (e.g., Figure 4).

Another interesting application of the data would be to triangulate or complement them with other data sources. International environmental meetings are often covered by several publications or media outlets. Notably, ENB reports usually publish a daily summary of the previous day’s proceedings, and the UN webcasts formal negotiation sessions live and makes recordings of the plenary sessions available to the public. The MARIPOLDATAbase refers to both data sources; it facilitates comparing and complementing by providing a link to the relevant ENB report of the day with each observation (enb variable) and a link to UN webcast recordings (where applicable, i.e., formal plenaries and working groups). By using the statements’ time stamp, recorded passages can be found quickly. Additionally, draft documents, other textual documents, and interviews can help to complement research (Langlet et al. 2023b). Another important source of information for any scholar aiming to trace the negotiations and find detailed information on provisions and positions is the uploaded text proposals. While they are not part of this database owing to its focus on the actual negotiations, that is, delegates’ speech acts, they constitute a very important source that can complement this database. Many—however, not all—text proposals can be found on the BBNJ dedicated UN website.¹⁵ The United Nations Department for Oceans and the Law of the Sea (UNDOALOS) Secretariat offered a number of different tools and ways to submit and receive textual proposals, such as PaperSmart, which was used only during the first three IGCs and later on provided QR codes through which delegates could download text proposals on their devices and compilations of all submitted text proposals.

This database also offers an entry into comparing the BBNJ negotiations with other negotiating fora, especially in the environmental field. For example, it could connect to research on the ongoing negotiations for a global plastics treaty negotiated under the UN Environmental Programme mandate. One way to connect and compare research on different international environmental negotiations and policymaking fora could be through data repositories such as manydata,¹⁶ which serves as a portal for different data sets on environmental governance that are related to one another through a standardized naming and identification system. Another way to use the MARIPOLDATAbase to triangulate different data sources could be by connecting it to social media data. For example, BBNJ topics were discussed and shared intensively on Twitter (now X). From Twitter data on BBNJ discussions, Langlet et al. (2023a) created a data set containing more than 40,000 tweets with the hashtag #BBNJ, which is now publicly available and can be used for research. The observational data from the MARIPOLDATAbase can, for example, be used to guide explorations into the Twitter data set or other openly accessible data sets.

When conducting research on global environmental agreement-making during multilateral conferences, it is essential to reflect on the ethical and practical conditions of accessing negotiation sites, collecting data, and using them for various types of analysis (Vadrot and Hughes 2023). We applied the highest ethical standards to the research design while the on-site data collection complied with the rules enacted by the ERC.¹⁷ In practice, this meant that we informed the UNDOALOS Secretariat (at the UN Division for Ocean Affairs and the Law of the Sea) about the research, the data that we would collect, and the anonymization and pseudonymization methods that we would apply.

Several steps were undertaken to respect the personal information of participants and the confidentiality of negotiation settings. All individual names were replaced by a “[removed for anonymisation]” placeholder. Also, all actor names linked to exchanges of views and statements recorded in informal negotiation settings (intersessional work, online dialogues, and informal informals) were pseudonymized, both in the “actor” column and in the content of the observations using a combination of rule-based editing and manual inspection. This careful consideration is driven by an ethical position that emphasizes the researcher’s respect toward the confidentiality of spaces and rooms, sometimes limiting the possibility to use certain data for research. A fully nonanonymized version of the database is available for research purposes upon request after entering a data-sharing agreement with the authors of this article.

While pseudonymization of actor names in more than 20,000 observations may be painful from a researcher’s point of view, it is—in our eyes—necessary to ensure trust in multilateral negotiations. State representatives often need to reach agreement over contested sensitive issues amid diverging state and nonstate interests. During such processes, trust among negotiators and observers is of the utmost importance (Vadrot et al. 2021), and to build trust, it is often necessary to provide safe, confidential spaces (Vadrot and Ruiz-Rodríguez 2022). When researchers receive access to such spaces and negotiation settings, they must also respect the anonymity rules attached to those spaces and a researcher’s access. Because successful conclusion of multilateral negotiations is important, oftentimes to secure peace or, in the context of the environment, to improve environmental governance, and delegates need to trust each other and the confidentiality of informal negotiation settings, we see the research interest as secondary after the adherence to confidentiality rules. This, in our view, also applies when the negotiations have officially concluded, because the implementation of the BBNJ agreement requires maintenance of trustful relationships among delegates, governments, and the secretariat; thus the conclusion of the BBNJ negotiations in this case only means the end of one intergovernmental encounter in a line of many more to come.

Global environmental negotiations are central sites of global environmental politics (GEP) research and the study of multilateral environmental agreements. They are of particular interest to scholars seeking to understand the processes behind the making of new law rather than the final outcomes. Broadening our understanding of global environmental meetings as organized social spaces (Langlet and Vadrot 2024a) and agreement-making sites (Hughes and Vadrot 2023) offers new insights into actor constellations, networks, power relations, forms of influence, and contestation shaping the global order. However, considering negotiations—such as the large-scale climate and biodiversity COPs involving tens of thousands of people—as field sites in their own right is a methodological challenge. By introducing the MARIPOLDATAbase, we propose a methodological shift in the study of global environmental negotiations echoing, first, recent attempts to elevate the meaning and empirical study of global environmental meetings (Corson et al. 2014; O’Neill and Haas 2019; Hughes et al. 2021; Hughes and Vadrot 2023) and, second, the principles advocating for improved findability, accessibility, and reuse of digital assets and data, such as the ‘FAIR Guiding Principles for scientific data management and stewardship’.¹⁸

The MARIPOLDATAbase is a reliable and transparent repository of observations on the BBNJ process. It includes systematic data on actors and alliances, statement length, agreement text, positions, networks, statements, concepts, and meeting formats. This publicly available data complements existing sources, such as ENB reports, social media records, and other empirical studies, improving and increasing the triangulation of data. We have designed the MARIPOLDATAbase to be both an open access repository available to all researchers and practitioners needing data on the BBNJ negotiations and a blueprint for future data collection, curation, and management in and beyond GEP research. As an open access repository, it aims to improve the reuse of data (by researchers both within and outside of our team) by increasing the findability and accessibility of data while at the same time paying close attention to ethical standards along the entire research life cycle from design (e.g., ethics approvals and data management plan) to data collection (e.g., consent forms, reflection on the positionality of field researchers), data curation and management (e.g., anonymization, pseudonymization), interpretation, and writing (e.g., authorship and acknowledgments). As such, the MARIPOLDATAbase acknowledges the need for slow scholarship counteracting the tendency toward “conference journalism,” rushing scholars into the premature interpretation of meeting dynamics and outcomes and the disclosure of sensitive information and data. As a blueprint for future data collection, curation, and management, it puts forward an innovated collective and collaborative field note–taking practice (see Vadrot et al. forthcoming) that might be valuable for scholars researching negotiation processes both in and beyond the field of global environmental politics; this would include diplomatic practices and UN meetings in areas such as trade, health, chemical pollution, and human rights. Our approach can be used in any setting where actors interact both with each other and with a legal text within a predefined meeting framework.

The MARIPOLDATAbase offers a diverse range of research opportunities, material for university teaching and training, and new possibilities for the future of treaty interpretation and development. Understanding the processes behind a specific provision, such as what actors were in favor of or against a specific provision and what terms were included to accommodate different state positions on monetary benefit sharing of marine genetic resources, may help future scholars and even diplomats involved in the future BBNJ COPs to anticipate conflicts and properly interpret the wording negotiators agreed on. Beyond the advancement of scholarly work related to global environmental negotiations broadly and specifically the BBNJ negotiations, this database may also support efforts to recreate the intentions and motivations of draft authors post hoc when interpreting the actual text. So-called travaux préparatoires are a well-established resource in international law and have commonly been highlighted as a valuable resource for treaty interpretation (Gardiner 2015) under Article 32 of the Vienna Convention on the Law of Treaties of 1969.¹⁹ The MARIPOLDATAbase could thus be used as a “supplementary means of interpretation” (Vienna Convention, Article 32) not only for the COP of the BBNJ Agreement but possibly also for discussion arising under the dispute mechanism before ITLOS. For example, the travaux préparatoires of the Refugee Convention of 1951²⁰ have become a widely used resource to give meaning to the text of the convention. The forthcoming BBNJ COP will still have to operationalize and interpret many provisions—for which underlying reasoning and arguments may be needed to understand the intent.

The BBNJ agreement has been shaped by historically embedded power relations and interests that are shaped by, reflect, and have the potential to remake or transform the intertwined global order of social, political, and economic relations. It is part of a complex web of international environmental agreements, diplomatic protocols, routines, and UN procedures that have the tendency to black-box the power struggles and global inequalities that perpetuate agreement-making, fostering global sustainability transformations and environmental justice. As scholars participating in and studying related actors, sites, and processes, we have attempted to elevate the reliability of negotiation data by constructing our MARIPOLDATAbase in a reflexive, reliable, and transparent manner, without neglecting the sensitivity of multilateral negotiations and their critical importance to making of global order.

The MARIPOLDATAbase as well as the technical document are available at: https://doi.org/10.11587/0XXZ0V. The data set should be cited as follows: Langlet, Arne, Alice B. M. Vadrot, Simon Fellinger, Paul Dunshirn, Silvia C. Ruiz-Rodríguez, and Ina Tessnow-von Wysocki. 2024. MARIPOLDATAbase (SUF Edition). https://doi.org/10.11587/0XXZ0V, AUSSDA, V1. The underlying code in R to create the figures displayed in this publication can be found and accessed here: https://github.com/ArneLanglet/MARIPOLDATAbase-publication-figures.

This work was supported by the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (grant 804599). The research presented is part of the ERC project MARIPOLDATA (https://www.maripoldata.eu/), led by Alice Vadrot. The authors give special thanks to the International Studies Association (ISA), UNDOALOS, the BBNJ Secretariat, and the organizers of the High Seas Treaty Dialogues for granting access to the research team to the different BBNJ negotiation formats. We thank our advisory network, colleagues, and critical friends who have supported our research since 2018, especially Gabriele Goettsche-Wanli, Thomas Loidl, and Philipp Bittner. We also thank the Austrian Social Sciences Data Archive (AUSSDA) and, in particular, Iris Butzlaff and Lisa Hirsch for their work on the data set.