Barcamp Open Science 2026: Navigating Challenging Times

In light of political changes that threaten democracy and academic freedom – currently visible under for instance the Trump administration but also looming in Germany, like in Saxony-Anhalt, where the AfD seeks to influence what should be researched – this session explored how Open Science can become more resilient and, in turn, help strengthen democracy. Saxony-Anhalt served as a case in point: a research-weak state with the lowest share of GDP spent on research funding among German states and generally low public support for science. Participants discussed a state-wide Open Science initiative that is already planned, though state funding is currently blocked ahead of the September state elections (Landtagswahlen). Networking with comparable structures (Open Research Offices in Berlin, Brandenburg, Saarland, NRW) and research networks such as Max Planck was also proposed.

Two core themes emerged: funding for Open Science and building public trust in science. On funding, ideas included joint archives, NFDI, the European Open Science Cloud, and TIB’s decentralized, BitTorrent-based SHIELD project. On trust-building, proposals centered on bringing scientists closer to the public – especially to rural areas, which make up much of Saxony-Anhalt – alongside citizen science approaches and combining scientific literacy with political empowerment to counter science skepticism and help people distinguish fact from misinformation.

Strategically, university leadership was encouraged to proactively engage with politicians, and institutions were seen as potential voices against extremism and restrictions on academic freedom. Underused instruments such as UN treaties (notably Article 15 of the ICESCR, PDF) and the UNESCO Recommendation on Open Science were also mentioned. An open question remained: should the Open Science community become more political – and if so, how, without merely “preaching to the choir”?

Participants reflected on how institutions, funders, and journals are trying to encourage Open Science, and why these efforts still often fail to change everyday practice.

Many “carrots”, such as recognition and career credit, were still seen as largely hypothetical because their promised benefits rarely materialise. For example, Open Science may be discussed during tenure-track hiring, but carries little weight in final decisions. Participants also noted that disciplines and institutions are at very different stages of Open Science adoption, meaning that no single incentive model will work everywhere.

At the same time, “sticks” such as funder mandates, legal obligations, and institutional expectations are often weakly enforced. Participants pointed to funder requirements for data management plans, which are rarely followed up or checked for compliance. Participants from “sticky” institutions with top-down legal obligations also observed that staff still look for ways to avoid publishing data because they fear that imperfections will become visible.

The discussion concluded that both carrots and sticks remain necessary, but neither will be sufficient without broader cultural change. Participants highlighted that the pace of science may need to slow down rather than continually adding new expectations. To address “error culture”, they suggested promoting the idea, in the spirit of George Box, that “all data are imperfect, but some data are useful”. Finally, participants argued that researchers should be more confident in badging their work as Open Science and framing it as a fundamental component of good science.

The session asked what happens to the research record if synthetic or AI-treated data quietly enters it undeclared by humans and agents alike. Participants worried about a growing flood of “AI slop” contaminating research datasets, so that future analyses and models are built on undisclosed machine-generated content. The group also noted that AI-introduced bias extends beyond text generation into literature search, data analysis and even study design.

They agreed that disclosure matters, but setting a threshold for when and how AI use must be declared is hard. AI-computed or synthetic datasets already exist in fields like genomics, and enforcement is messy. Two concrete ideas emerged: research instruments such as sequencers and microscopes could cryptographically sign or watermark the data they produce, marking it as AI-free at the source. And content-derived identifiers like the ISCC (International Standard Content Code, ISO 24138) could help persistently identify digital content and carry origin declarations across systems.

Instead of chasing a precise boundary, the group converged on a pragmatic principle: be honest about what you used, and make it explicit. A short reflective statement on risk of bias was seen as more meaningful than a bare “AI was used” checkbox. Simply archiving prompts was viewed as limited, given that the same prompt can yield different outputs, but documenting which model was used can support bias assessment. Ideally, tools like electronic lab notebooks capture such provenance automatically, making room for reflection as the core of responsible AI documentation.

This session focused on providing a literature dataset including empirical research on Open Science practices and discussing its use for informing daily Open Science advocacy work or further science studies. In 2020, a group of scholars from the DACH-region started to annotate published research from 2000 to 2020 to capture Open Science practices, methods, disciplines, and target groups, as well as countries of studies, to gain insights in international developments of Open Science. The literature collection and mapping will be continued and expanded for the following years.

The question to the participants, who were Open Science advocates and researchers, was hence, what information on Open Science practices they need in their daily work and how the team could take these needs into account to enrich the dataset. One recommendation was that use cases or comparable good practices of Open science and its operationalisation would be a major contribution. Furthermore, the systematic survey could reveal more about trends and emerging Open Science topics that matter for science management.

Another topic centered around meta-science questions and methods: For instance, it is yet not clear whether Open Science is emerging as an independent field of study. Additionally, the team will have to check for opportunities to include the growing annotated dataset into larger Knowledge Graphs projects, to be explorable and usable for meta-scientists and Open Science practitioners alike.

Various types of events exist where Open Science is promoted, discussed, and advanced – such as the Barcamp. People who attend these gatherings are often already familiar with the topic and in favor of pushing the Open Science agenda. However, a (literally) critical mass is needed to achieve the cultural changes required for Open Science to become the norm rather than the exception.

So how do we involve academics and other stakeholders who are not yet interested in, or may even be critical of, Open Science? In this session, we identified some relevant target audiences and then decided to further focus on university students training to become teachers. We employed an empathy mapping approach, visualizing what our target audience does, hears/sees, thinks and feels. This was then the baseline for discussions around how to practically engage this group.

We worked out that prospective teachers do not get a lot of input on didactics, are under a lot of pressure and also get confronted with current political developments. Engagement of this target audience should center on offers that sound attractive and does not seem like additional workload. We speculated that a promising approach would be to offer them practical tools for scientific literacy instruction and for demonstrating the benefits of science to students. Other highly relevant topics may be Open Educational Resources and the debate of “knowledge (and school books) as a public good” (which led to the Barcamp-Session Open Educational Resources for schools).

In this year’s edition of this recurring session (edition 4 already!) we were a small group of people mainly from the natural sciences. The goal of an Open Science success stories session is to round up experiences and positive examples of successes that we or others achieved by embracing Open Science practices, tools and patterns of thought. We hope these success stories inspire and can serve as good practice examples to help get more people on board in doing Open Science.

Here are three of the success stories shared this year:

1. The Human genome project as an example of an Open Access project that started in 1990 and was a huge collaborative effort to sequence the human genome with the data being made available together with the publications.
2. The trailblazing life science journal “eLife” that trained researchers through its ambassador program to apply Open Science principles in their work and communities, an effort that led to career boosting opportunities for many of the ambassadors and continues to shape the way they approach their work.
3. Increased visibility of a researcher and successfully applied Open Science principles that made data publication easier through coverage of a project in a newspaper.

Our key outcomes of this session were that through trailblazers from the past that demanded and applied Open Science principles we can now see the positive effects and reap the benefits such as easier data publication and increased visibility.

The session examined the frequent replacement of school books, which are mostly proprietary and rarely available in digital formats. Participants discussed why this happens and how the situation could be improved. Although not a core Open Science topic, the issue involves similar stakeholders, such as publishers. The group envisioned school book content evolving into Open Educational Resources (OER), maintained by a community and potentially supported by commercial entities. These resources would be freely accessible online under open licenses, easily adaptable, translatable, and continuously developed over time. They could be printed on demand. The main obstacle seems to be the financial incentives for publishers, which currently result in a suboptimal experience for students and the public. The session also highlighted the movement advocating for OER in education and the upcoming OERCamp at the end of August offering a valuable opportunity to connect and advance the cause.

The starting point for this session was the recurring topic of data rescue and the resilience of research data infrastructures. Efforts to create a decentralized infrastructure for cooperatively safeguarding research outputs, for instance using BitTorrent, introduce new challenges for research data governance. When the original repository where a research output was published goes down, the data governance mechanisms associated with it may also become unavailable.

The session therefore asked which situations may pose governance challenges in a decentralized, openly networked infrastructure. In a small group, we identified several scenarios: publishing a new version of a dataset, requesting corrections to metadata and the replicated metadata files, for example, if a researcher changes their name; or even temporarily restricting access to research outputs that may endanger their creators in situations of political repression.