Imagining sustainable futures for the high seas by combining the power of computation and narrative

The paper addresses how to envision sustainable futures for the high seas amid rapid industrial expansion, biodiversity loss, unequal resource access, and fragmented governance in areas beyond national jurisdiction. It posits that transformative change requires shifts in paradigms, mind-sets, and human–ocean relationships. The authors argue that imaginative, plural, and empirically grounded futures can catalyze social-ecological transformations, expand the perceived option space under uncertainty, and inform ongoing policy processes (e.g., BBNJ negotiations). By combining computation and narrative, the study seeks to produce radical yet plausible visions that challenge current trajectories, foster collective imaginaries, and support decision-making.

The paper situates its work within literature on the ‘blue acceleration’ and its consequences (biodiversity decline, overfishing, inequitable benefit sharing), emergent ocean industries (geo-engineering, marine genetic resources, deep-sea mining), climate change impacts, and the governance complexity of ABNJ. It draws on scholarship on transformations, leverage points, and sociotechnical imaginaries, highlighting the role of imagination and plural values in shaping desirable futures. Prior work shows creative, story-based visions (including science fiction prototyping) can motivate, broaden anticipation capacities, and communicate complex systems. The authors connect to futures methods (topic modeling for evidence synthesis; structured futuring; future wheels; cross-impact; three horizons; cultural iceberg), and argue for art–science collaborations to address the ‘imagination gap’.

Overall framework: The study blended computational text analysis (topic modeling) with a structured futuring approach to generate science-fiction stories, followed by expert interviews to assess links to current realities and potential to foster new imaginaries. Data collection: Using Scopus, the authors queried terms related to the high seas, governance, and futures, restricted to post-1982 (post-UNCLOS) publications. The search returned 8,846 results; applying the cutoff yielded 8,580 abstracts for analysis. Topic modeling: Texts were cleaned (tidy structure, stopword removal) and converted to a document-term matrix. Multiple algorithms (VEM0, VEM_fixed, Gibbs, CTM) were compared on entropy, log-likelihood, and perplexity (where applicable). The Collapsed Gibbs Sampler was selected for its performance. Models with varying topic numbers were tested; although statistical optima suggested many topics (80–100), the study chose 25 topics to balance interpretability and parsimony. Topics were visualized using LDAvis; lambda (relevance) was set to 0.6 to improve interpretability by de-emphasizing corpus-common terms. Sixteen of the 25 topics were retained for scenario inputs (others were off-scope or uninterpretable). Structured futuring: Steps included (1) summarizing topic keywords into labels; (2) clustering four diverse topics per scenario using intertopic distance (ensuring STEEP diversity); (3) constructing future wheels (first- and second-order consequences); (4) cross-impact matrices to explore interactions, synergies, contradictions, and feedbacks; (5) mapping outcomes onto the Three Horizons framework to stage transitions; and (6) elaborating socio-cultural dimensions via the cultural iceberg model. Elements were sometimes ‘pushed toward ridiculousness’ to overcome linearity and blind spots. The outputs informed iterative writing of four science-fiction stories with accompanying commissioned artwork. Interviews: Eight semi-structured interviews with international high seas experts (six in-person during a Cape Town workshop, two via Zoom; ~20 minutes each, in English) explored whether story elements mapped to current realities/evidence and assessed the value of story-based visions for research, communication, and policy. Interviews were recorded and transcribed (not publicly available per consent). Deductive coding yielded four categories: (1) links between story elements and current realities, (2) benefits/potential of creative methods, (3) use cases/applications, (4) limitations/critiques.

- Topic model: 8,580 abstracts analyzed produced 25 topics (16 used for scenarios). LDAvis aided interpretation (lambda=0.6). These topics spanned concrete themes (e.g., conservation, AUVs, hydrothermal vents) and more abstract clusters (e.g., shipping/trade with surprising terms), enriching creative inputs. - Four story-based visions: The Ocean Uprising; Deep Connections; Kairei City; Myopia. They depict contrasting futures (technocratic seabed privatization; deep-sea societal adaptation; a high-seas nation-state with mobile infrastructures and rights of nature; total ocean transparency with ubiquitous sensing and enforcement challenges). - Linkage to present realities: Expert interviews confirmed numerous elements as extrapolations of current trends or present discourse/evidence (e.g., privatization/industry stewardship analogues like FIPs; deep-sea habitation concepts; floating cities/sea-steading; improved/empowered RFMOs; transparent oceans via open data and pervasive sensors; IUU tactics like disabling AIS/VMS; rapid fish evolution under pressure). Tables 1–4 provide representative quotes. - Transformative potential of narratives: Interviewees emphasized stories and art–science collaborations as powerful for communication, outreach, emotional engagement, and widening participation. They reported that the visions broaden thinking, encourage experimentation beyond current frames, and could support planning, horizon scanning, education, and policy deliberations by legitimizing unconventional ideas and expanding perceived option space. - Quantitative details: 8,580 abstracts; 25 topics (16 used); 4 visions with visual artworks; 8 expert interviews (~20 min each).

By combining computational synthesis (topic modeling) with structured futuring and narrative, the study generated radical yet evidence-grounded visions that audiences found traceable to current realities. This dual grounding—quantitative literature structures informing imaginative storytelling—addresses the ‘imagination gap’ by expanding what is considered plausible and discussable. The interviews suggest that such visions can open transformative spaces, enabling emotional connection to distant ocean spaces, challenging assumptions, and supporting the co-creation of new high-seas imaginaries that could inform policy (e.g., BBNJ negotiations). The findings demonstrate that futures work benefits from plural perspectives and narrative forms to motivate, direct attention, and potentially reshape socio-technical norms and institutions toward sustainable and just ocean governance.

The study presents a methodology that fuses computational topic modeling with structured futuring and science-fiction storytelling to craft radical but scientifically anchored visions for the high seas. It contributes a set of four narratives and associated artworks that link to present evidence while exploring diverse, non-linear trajectories, thereby creating spaces for reflection, dialogue, and potential transformation of human–ocean relationships. Future research should incorporate participatory co-creation to diversify perspectives, involve artists throughout the process, empirically test the effects of engaging with story-based visions, expand beyond English-language literature, and broaden audiences to assess resonance and utility across communities and decision-making contexts.

- Corpus and language scope: Only English-language abstracts from Scopus were analyzed, potentially biasing topic coverage and perspectives. - Model interpretability choices: Selecting 25 topics (for parsimony and interpretability) trades off with statistical optimality suggested by higher topic counts. - Scenario selection: 16 topics were chosen subjectively for visioning; discarding others may limit thematic diversity. - Interview sample: Small (n=8) and comprised participants familiar with creative futuring methods, which may bias perceptions toward receptivity; findings may not generalize to all high-seas experts or stakeholders. - Non-public interview data: Recordings/transcripts are not publicly available, limiting external verification. - Transformations framing: Authors caution against treating transformations as inevitable or universally beneficial; unintended consequences in teleconnected social-ecological systems remain a concern.