Redefining Hallin and Mancini’s Media System: Cross‐Border Investigative Networks in Europe

,


Introduction
Hallin and Mancini's book Comparing Media Systems: Three Models of Media and Politics has, since its publication, been recognised as one of the most significant contributions to communication studies (Hallin & Mancini, 2004).The conceptual framework designed by Hallin and Mancini facilitated the identification of similarities and differences between media organisations, journalism cultures, and professional practices, giving rise to numerous taxonomic studies in this field.Hallin andMancini (2012, 2017) revisited their previous work in 2012 and 2017, incorporating updates to adapt it to the global digital context (Humprecht et al., 2022;Liu et al., 2020).The appearance of terms such as "hybridisation" (Hallin et al., 2023) and "volatility" (Mancini, 2015), ensured the continued validity of the model defined in 2004 in a convergent environment open to the innovations introduced by digitalisation (Flew & Waisbord, 2015).
These references to the global digital paradigm inform the framework that underpins this study.The fluidity of communication exchanges resulting from a phenomenon of internationalisation-associated with networks that connect journalists across national borders-requires us to broaden the scope of Hallin and Mancini's model.Hallin (2020) himself has recognised the impact of transnationalisation and how this phenomenon, accelerated by the internet, has reinforced the idea that media systems are not closed or autonomous models, opening up the possibility of expanding and refining the original categories defined in 2004.This idea has also been supported by Kraidy (2011), who, in his research on the pan-Arabic media space, analyses how media corporations operating across borders incorporate conceptual innovations.As examples, he points to transnational parallelism and a broader conception of the roles played by professionals.
One characteristic example of these transnational media initiatives can be found in the networks of journalists who operate between countries and collaborate on the investigation of major global news stories.These networks have grown exponentially in the last few decades (Krüger et al., 2019), and yet they have received barely any scholarly attention in proposed revisions to the work of Hallin and Mancini.It seems clear that the "ideal models" (Hallin & Mancini, 2017, p. 159) defined in 2004 could be enriched by empirical studies like this one, given that cross-border investigative journalism (CBIJ) escapes state-centric rhetoric (Couldry & Hepp, 2009), moving in an "in-between space" (Hellmueller & Berglez, 2022, p. 15) where traditions, practices, and narratives of different models are all combined.
Given the above considerations, this article analyses the development of the idea of CBIJ in Europe and its contribution to an adaptation of the classical model of media systems to a transnational phenomenon, incorporating new values related to aspects such as the financial models that support these initiatives, their connections to political structures, the topics explored, and professional practices.In this way, the spectrum of the original dimensions of the press market, political parallelism, and journalistic professionalism can be broadened in consonance with other studies that have updated the scope of Hallin and Mancini's model over the last two decades.This study thus seeks to test the following hypotheses: H1: CBIJ prioritises non-profit organisations supported by foundations and reduces the importance of monetizable elements as it aims for an impact distinct from the financial gain expected in the commercial business model.H2: CBIJ transforms political parallelism into transnational parallelism to facilitate Europe's visibility as a priority topic and to cultivate a supranational consensus on basic human rights in the EU.H3: CBIJ encourages a higher level of internal and external autonomy for journalists in these networks to foster professional practices that can overcome the limitations of the competitive models of conventional media.
To test these hypotheses, this study examines the role of Dataharvest (the European Investigative Journalism Conference) in shaping CBIJ.
Dataharvest has become a key player in collaborative cross-border journalism.This organisation has been consolidated as a European extension of highly institutionalised investigative networks in the US, such as Investigative Reporters and Editors, the Global Investigative Journalism Network, and the International Consortium of Investigative Journalists, organisations that have been responsible for shaping the narrative on investigative journalism since the 1970s.Candea (2020) and Houston (2016) highlight the importance of these events in formalising and promoting a particular way of understanding and doing journalism.Their organisers choose the session participants, the topics to be addressed, and the potential areas of discussion and influence.Lampel and Meyer (2008) also explore the significant value of analysing events like these to identify their influence on the evolution of certain professional fields.From the perspective of organisational sociology, Haug (2013, pp. 712-713) points out how these spaces establish a principle of order through the relationship of trust they create between participants and organisers.The scale of this consensus supports the idea of a conceptualisation process operating at these events, where opinions are seen through the lens of the cooperative and a group notion is constructed on certain issues.The creation of these interpretative communities gives those who belong to them a consciousness and identity as members of a specific professional field (Lampel & Meyer, 2008, p. 1027).
Adopting the term used by Schüßler et al. (2015, p. 169), the annual Dataharvest conference could be described as a "mega-event," with a high degree of legitimacy for influencing the narrative on CBIJ.
The number of participants has increased exponentially over the years, from 35 in 2011 to more than 500 in its most recent edition.It also has an independent organisation (the Arena for Journalism in Europe) responsible for its organisation, and it is now considered a "global hub" for sharing resources, discussing experiences, and receiving mentoring (Heft, 2021, p. 460;Heft et al., 2019Heft et al., , p. 1187)).
Recognising this function, this study aims to identify the topics addressed at Dataharvest since 2014 and to determine the frequency and evolution of those topics over the years.This frequency and relationship analysis will establish a hierarchy of content to identify the topics that have dominated the conferences in discussions of the categories of funding, professional practices, thematic diversity, actors, and connections of these transnational networks.These categories are based on the conceptual framework described in Section 2.
The approach taken in this research connects it with the discursive processes of meta-journalism described by Carlson (2016), who highlights the importance of public expressions like the annual Dataharvest conferences for shaping an understanding of journalism as a cultural practice that is interrelated with the contextual (social, economic, technological) conditions of the time.
It is not the aim of this article to explore the extent to which these ideas have permeated the professional consciousness of conference attendees.However, the literature consulted does provide qualitative data on this, which will be compared against the findings of this study in Section 5.

A Conceptual Toolkit to Understand CBIJ
Collaborative investigative journalism has grown exponentially over the last two decades.As a result, scholarly research on this phenomenon has also increased.Some authors have taken an approach to CBIJ that focuses on the analysis of professional practices (Alfter, 2016;Alfter & Candea, 2019;Heft, 2021;Heft & Baack, 2022;Konow-Lund, 2019;Wuergler & Cancela, 2022), while others have highlighted its capacity to offer a solution to the current crisis in journalism by fostering synergies between media organisations facing financial difficulties (Michailidou & Trenz, 2023).The technological dimension has also been an important focal point (Bird & Candea, 2017;Bunce et al., 2018;Ng, 2021), with detailed descriptions of the digital technology that has accelerated this practice in contemporary society.
Integrating all these approaches, the theoretical framework for this study describes the categories of funding models, professional practices adopted in newsrooms, topics explored in reporting projects, and participants in these networks and their connections.As a starting point, this research adopts the description of CBIJ offered by Graves and Konieczna (2015), who define it as a form of "field repair" that adapts media practices to the complexities of the real world.Responding to journalists' dissatisfaction with their companies' limited resources (Heft, 2021, p. 462), CBIJ has become a "space" where professionals can break free from the limitations imposed by editorial lines and commercial interests.It thus enables them to take on a more ambitious, wide-ranging mission, reflected in the potential of teams working on global stories that are silenced in mainstream media (Alfter, 2021, pp. 219-220).In this way, CBIJ can be associated with the recovery of quality journalism (Coronel, as cited in Houston, 2021Houston, , p. 1094)), reclaiming basic values such as accountability, transparency, and the exposure of abuses of power.Indeed, various EU institutions have confirmed this association, specifically highlighting the positive effects that cross-border cooperation between investigative journalists has on the quality of the information provided to the public (European Commission, 2018, p. 30;European Parliament, 2018, pp. 9, 12).

Sustainable and Impactful Alternatives for CBIJ
The expansion of transnational journalist networks is intrinsically associated with the consolidation of non-profit organisations (Kaplan, 2013).The decline of the commercial model has given impetus to the search for financial alternatives in a saturated market that cannot guarantee enough resources for everyone (Clement et al., 2018;EUROPE Ltd & Media Consulting Group, 2014;Maness, 2013).Various formulas have been adopted, but the most prominent in quantitative terms has been funding from private and public foundations, representing 27 billion dollars since 2009 according to data from Media Impact Funder (https://mediaimpactfunders.org).
It is also important to assess the success and impact of CBIJ.In the profit-driven model, the organisation supports stories that readers display a willingness to pay for.In non-profit organisations, the impact is measured using different indicators, such as the visibility of salient topics and their capacity to generate conversation in other media, the creation of exclusive editorial value (Alfter, 2019, p. 5), and reputation, as reflected in awards and prizes.It is thus not a question of profits but of deliberative and substantive civic impact, with a decisive influence on the redefinition of public opinion and changes to political agendas (Hamilton, 2016, p. 93).

Norms and Practices in Intertwined Newsrooms
CBIJ adopts a formula of "intertwined newsrooms" (Buschow & Suhr, 2022, p. 295) with diverse journalism cultures integrated into hybrid networks.In such networks, resources are not distinguishable as belonging to a specific organisation, and innovative practices are developed that push the boundaries of conventional journalism (Mesquita & de-Lima-Santos, 2021, p. 548).These practices include the creation of international solidarity and shared ethical standards that prioritise trust and mutual assistance over competition (Hume & Abbot, 2017, p. 5).The synergies generated encourage professionals to take a shared interest in an issue and to coordinate materials, narratives, and publication dates.
Other notable benefits include the expansion of investigative capacities thanks to the "pooling of resources" (Konow-Lund, 2019, p. 103) and knowledge transfer between team members, as well as access to local experts and technical data skills.The success of these partnerships depends on a neutral editorial coordinator who can resolve the tensions that arise in projects of this scope (Sambrook et al., 2018, p. 29).

Systematised Thematic Diversity
The complexity of cross-border realities offers a wide range of possible topics reflecting considerable thematic diversity.However, the academic literature reveals a somewhat systematic dimension to these projects (Hamilton, 2016, p. 62), with political actors, corporations, and criminal organisations implicated in global stories of embezzlement and mismanagement.
In Europe, these networks address common issues on the supranational level (Grill & Boomgarden, 2017), with special attention to corruption and EU funding, but without ignoring other topics of special significance such as the environment, healthcare, the rise of the far right, or care for vulnerable groups.Archetti (2019) proposes a relational approach to these networks with a fluid map of exchanges between actors.This not only includes news industry professionals (editors, journalists, photographers, fact-checkers, etc.) and the sources they work with but also covers other professionals such as data scientists, coders, activists, intermediary organisations, etc.The connections between them are established mainly through the shared technology platforms (Bird & Candea, 2017;Bunce et al., 2018) that have been responsible for accelerating the development of these kinds of practices in a connected society (Carson, 2020;Gearing, 2016).

Connections
It is also important to consider factors such as the size and duration of the network, the integration of organisational factors, the institutionalisation of roles and tasks, and discussion and decision-making processes (Houston, 2021;Jenkins & Graves, 2022).These give rise to somewhat hierarchical and centralised structures, with high-or low-degree collaborations (Heft et al., 2019(Heft et al., , p. 1189)).The workflows in these collaborations may be vertical, proposed by editors or coordinators, or the product of the pioneering practices of individual journalists seeking to introduce a differentiating factor into their work environments (Hepp & Loosen, 2021, p. 590).

Methodological Design
A quantitative methodology has been adopted for this study based on automated content analysis employing text-mining techniques.A total of 1,015 documents (87,589 words) containing the abstracts for the sessions held at the conferences from 2014 to 2023 were analysed.The texts were downloaded from the Dataharvest website (https://dataharvest.eu) and subjected to a first-level manual clean-up prior to indexing, information extraction, classification, and analysis.
First of all, a basic frequency analysis was conducted to obtain a count of the essential units of language and their evolution over the period studied.Various grammatically related terms and synonyms were considered, after eliminating stopwords.The bag-of-words frequency model was applied for untransformed counts and the term frequency-inverse document frequency measure was used with a logarithmic reduction to highlight words that were less common but still significant in the corpus.
Secondly, n-grams were used to detect adjacent terms and find meaningful associations.Bigrams and trigrams insert the words into their context for a better understanding of the relational codes identified in the Dataharvest sessions.
Finally, a cluster analysis was conducted using transformers applied to document vectors with sBERT.This natural language processing technique, developed using a pre-trained AI model, was adopted due to its capacity to perform semantic searches and understand the context of a word based on the words coming before or after it.The results obtained were grouped using k-means clustering, which shows relationships that are not apparent at first sight and identifies patterns that help categorise unstructured data into coherent groups with shared qualities.
Automated text analysis has been applied in two consecutive stages.The first was the extraction stage, where essential language units were isolated and quantified by applying the bag-of-words model and term frequency-inverse document frequency.The second was the classification stage, which involved the organisation of the terms into nominal categories based on the conceptualisation of CBIJ outlined in Section 2. The development of the analysis protocol began with an initial datasheet resulting from the theoretical review.After a preliminary superficial exploration of the corpus, some variables were recoded to fill in the gaps detected.
As reflected in Table 1, four main research categories were defined: funding model (RC1), professional practices and routines in collaborative newsrooms (RC2), topics explored in reporting projects (RC3), and participants in these networks and their connections (RC4).The indicators selected were identified as potentially useful for clarifying how the original variables of the press market, political parallelism, and journalistic professionalism have expanded in the context of globalisation and convergence.
Two levels were specified within each category: the first with a macro variable, and the second with the keywords that would facilitate the semantic allocation of the language units obtained in the count to the level 1 variables.The complete list of keywords for the second level can be consulted in the Supplementary File.After an initial automated allocation using an index of coincidence, the textual context of these units was reviewed manually in the original documents with the aim of disambiguating words that could be classified into multiple categories.

Frequency Analysis
Figure 1 presents the 20 most frequently used terms.The word with the most appearances is "data" (1,079), well ahead of "journalists" (519), "journalism" (408), "European" (291), "EU" (289), "investigative" (271), "work" (242), and "Europe" (228).The combined repetitions of words related to the physical and symbolic supranational space ("European," "EU," and "Europe") place this concept in third place with 818 repetitions, behind the 927 combined repetitions of the generic "journalists" and "journalism."The rest of the content words in this first count refer to elements associated with professional practices (including "information" with 222 repetitions; "tools" with 184; "research" with 180; and "project" with 173), the transnational nature of the teams ("crossborder" with 195) and the type of space being covered by the news ("public" with 187).
The analysis was then expanded to include words used 20 times or more to increase the corpus and map a more complex semantic space related to CBIJ.A total of 492 items were obtained and grouped according to the descriptors defined in the methodology section.The table with the full classification is included in the Supplementary File.
The results reveal that the issues dealt with the most at Dataharvest conferences are those referring to professional practices (RC2), with 192 words (11,967 total repetitions).Next is the category of connections (RC4), with a lower number of more frequently repeated units (77 words and 4,827 repetitions), followed by the category of topics investigated (RC3; 91 words and 4,168 repetitions), and finally funding models (RC1; eight words and 251 repetitions).In RC1, words related to grants/donations from foundations predominate.This category also includes the specificity of the non-profit model, but other words reflecting the diversity of funding sources, such as crowdfunding (eight repetitions), prizes with a monetary component, and sponsorships (three repetitions), are left out of this list.References to the sustainability (five repetitions) and monetisation (three repetitions) of these projects and to entrepreneurship (14 repetitions) are also excluded.
In addition to skills associated with data processing and the use of specific software, the category of professional practices (RC2) includes terms related to information sharing, leaks, whistleblowing, and storytelling techniques.The study identified a wide range of words related to the day-to-day activities of journalists: "investigate," "explain," "find," "understand," "look for," "discuss," "report," "publish," etc. Freedom of information is also mentioned, and there are references to the risks and threats that professionals in these networks are exposed to, with special attention to online security.These issues, along with related legislation (which is also included in this ≥20 frequency list), form part of the macro-context in which reporters work.
In relation to the essential principles that define the profession, "transparency" (47) and "accountability" ( 19) have replaced traditional terms such as "objectivity," which was only mentioned once in a decade, or "watchdog," with only five mentions.References to ethical behaviour are also in a secondary position, with a frequency rate of six.However, there is a notable number of references to the independent status of these journalists (28 mentions), in constant dialogue with the pressure groups and stakeholders with whom these networks can collaborate.This results in references to the editorial "credibility" of reporting projects, with a lower frequency of seven mentions.
The terms identified reflect a standardised professional with no collaborative dynamics in transnational networks mentioned with significant frequency.For example, the notion of the "editor in chief," highlighted in the academic literature as an essential role, was mentioned only four times.
The spectrum of topics dealt with by cross-border teams expands beyond the concept of "Europe," which holds all the top positions in RC3, to explore other areas such as news coverage of transnational corporations.
Special attention is given to the agrifood and real estate industries, as well as the financial sector, corruption scandals, abuses of power and criminal activity, the power of lobby groups, public tenders, tax havens, and tax evasion (with the "Panama Papers" and the "Pandora Papers" as paradigmatic cases).Other issues emphasised are basic human rights, the climate crisis, water access, healthcare, and labour issues.
RC4 can be mapped conceptually on the basis of terms that suggest community and group relations: "us," "share," "networks," "join," "communities," "colleagues," "group," etc. References to the different scales of these networks (local, regional, national, and global) are included among the most frequently repeated words, along with mentions of the online space as a facilitator of connections free of geographical limitations.Under the technology descriptor, open-source culture is also significant, as are news sources.
In the frequency analysis by year, it was found that the words "data," "journalists," "journalism," "European," "investigative," "research," and "crossborder" appear in all years, as can be seen in Figure 2.  The words "corruption" and "media" are prominent in 2015, as is the word "freedom."The word "security" appears in 2017 with 30 repetitions (it would not do so again until 2022), along with terms associated with news coverage of corruption and money."Housing" and "local" are consolidated in 2019, even to the point of constituting a section of their own."Climate" takes fourth place in 2020 (it appears as early as 2016, but with a lower frequency rate), also with its does the same in 2021 and "work" in both 2022 and 2023.The terms "labour" and "people" gain prominence in 2023, with references to European legislation and the exploitation of workers in certain sectors and countries.These frequency lists broken down by year do not provide any distinguishing information other than that indicated above related to the emergence of certain topics associated with social, political, and economic issues that were prominent in Europe at specific times.

N-Gram Analysis
Table 2 presents the bigrams with 25 or more repetitions.Once again, "data" has a prominent presence, with four bigrams linking the concept to transnational networks.The source of data is also an important question, especially given the consideration of open government data as a priority resource for investigations of this kind.
The pair of bigrams referring to the transnational scale framework are worth noting, with direct references to professionals operating "across Europe" and "across borders."References to the right of free access to information also have a prominent place in this list.Although freedom of information has not had a section of its own since it was added to the Dataharvest program in 2015, there are constant references to it using the terms "freedom of information" and "wobbing" to highlight the essential nature of this right.Equally notable are references to recurring news topics for this type of investigative journalism, such as organised crime and corruption, as well as security issues, with special attention to threats in the digital space.

Discussion and Conclusions
Journalist networks operating global stories have not been the object of a systematic academic review in proposals to update the media system model proposed by Hallin and Mancini.The exploratory study offered here contributes to the academic literature on this phenomenon, revealing how CBIJ has been conceptualised at Dataharvest and how it has broadened the comprehension of the media system described in 2004 and its subsequent revisions over the past two decades.To conclude this article, the main findings of the analysis are summarised below, contextualised with the academic literature on the subject.
In the case of the business model (H1), Dataharvest displays an evident interest in private foundations (Adessium and Rudolf Augstein Stiftung sponsored the first editions of the conference, for example) as a preferred funding option for CBIJ, limiting the monetising potential of these networks by relying on non-profit organisations.On this point, our results are highly consistent with other studies that signal the decisive role of foundations (e.g., Padania, 2019).The sections dedicated to the funding for this journalism underplay the importance of diversifying the revenue sources to achieve a more entrepreneurial model, like one based on collaborations with consolidated mainstream media organisations, for example.This would free the non-profit entities from dependence on donors and their interest in shaping public opinion by thematising the agendas of the organisations they support (Birnbauer, 2019, p. 177).
This last point has led Browne (2010) to point out the elitist nature of this kind of journalism, suggesting that it is consumed not so much by large transnational audiences as by small groups with notable decision-making power at the institutional level.In this respect, it is worth noting that references to audiences and the creation of strong connections with them using different strategies (such as gamification or interactive content) are not very frequent at the Dataharvest conferences.This reality raises questions about the continued relevance of the press market dimension as originally defined by Hallin andMancini (as cited in Brüggemann et al., 2014, p. 1040; "how far the press reaches out to a broader audience"), in light of the addition of new strategies to a form of journalism aimed at social groups that are much smaller but have a big impact on setting the public agenda.
This in turn raises the question of the content reported in CBIJ (H2) and the creation of a symbolic supranational space where news outputs are offered to multiple countries in a context of mutual understanding (Hellmueller & Berglez, 2022, p. 11).The results of this study reveal the predominance of Europe as a prioritised topic, confirming Flew and Waisbord's (2015, p. 626) , 2014, p. 29).This thematic predominance has, in fact, been the subject of criticism (Schiffrin, 2017), as certain topics are highlighted in a way that can limit the range of stories being investigated by transnational networks, which end up offering the same data, sources, and narratives.
Concerning the levels of journalist autonomy within networks (H3), Dataharvest conferences favour integration into consolidated structures, a phenomenon consistent with the findings of this study in relation to the prominence of high-level organisations (Heft & Baack, 2022, p. 2341) (Heravi & Lorenz, 2020, p. 36).For example, the recent global investigations carried out by consortia of journalists have been intimately linked to the mass leaks of huge volumes of data such as the "Panama Papers" and the "Pandora Papers."These two projects and their work techniques are given special attention at Dataharvest.This fascination with data, however, is not identified as positive in all the literature reviewed.Some authors criticise what they see as the imposition of a kind of "feudalism" due to the power of platforms and big tech over the data used.Candea (2020) refers to the hegemony of companies such as Google and Meta, pointing to the establishment of monolithic structures with clear imbalances arising from the lack of real control over the data by the team of professionals using them.Ownership and free management of data, as essential raw material for cross-border journalism networks, therefore needs to be added as a key indicator in order to define more precisely the autonomy of journalists, identified by Hallin and Mancini (2004) in their category of "journalistic professionalism," in the new convergent environment.
It is worth highlighting the scant attention Dataharvest has given to certain problems identified in the literature that have appeared only marginally in explorations of professional practices at the conferences.These include the questions of how to reduce competitiveness in multidisciplinary teams (Jenkins & Graves, 2022), how to manage increases in non-journalistic work (Ingram, 2019) and overlapping with daily routines.Other issues of a structural nature, such as the unconscious biases of journalists, conflicts of interests, ethical standards, or specialisation have also been largely ignored.
The preliminary results of this study contribute to a clearer understanding of CBIJ and its adaptation to an increasingly complex media environment with dialectic relationships between the global and the local, top media organisations and low-level independent networks, prioritised of the model established by Hallin and Mancini reveals that these networks do not fit within the parameters defined by these authors in 2004, as they have developed qualities of their own in relation to the business model applied, the transnational orientation of reporters, the topics addressed, and to a lesser extent, the professional practices adopted.As Archetti (2019Archetti ( , p. 2151) ) suggests, "labels don't apply," and a relational approach to this phenomenon is needed in order to understand its complexity.
The optimistic view of cross-border journalism as a collaboration for the common good (Martínez de la Serna, 2018) should not prevent future research from taking a critical approach to this phenomenon and the challenges it poses.In this respect, some authors argue that while collaboration is imposed as the dominant narrative, journalists operate within the cultural and professional context of their respective countries, thus undermining any genuine integration (Meyen, 2018;Michailidou & Trenz, 2023).This and other issues mentioned earlier (such as editorial interference by donors or the hegemony of data) require further study to determine the real impact of CBIJ on contemporary society.

Figure 2 .
Figure 2. Evolution (2014-2023) of the 10 most frequently repeated words.Note: The full results are provided in the Supplementary File.

Table 2 .
Bigrams with a frequency of 25 or more.Trigrams, as shown in Table3, are not as numerous, but significant relationships also appear in three-word sequences with 10 or more repetitions.The top five positions are held by variations on the names of two of the main transnational journalist organisations, the International Consortium of Investigative Journalists and the considerable distance behind the top five ("EU member states" with 12).Media and Communication• 2024 • Volume 12 • Article 7712

Table 3 .
Trigrams with a frequency of or more.36 argument that this kind of news coverage "challenge[s] the authority and decision-making capacities of nation-states."CBIJ is thus disengaged from national political structures because only in this way is it possible to report on global stories related to transnational actors.Contributing to this is transnational parallelism, as indicated in the second hypothesis for this research, which involves an alignment of the professionals in these networks with the basic operating principles of the EU.According to Ides Debruyne, managing director of Journalismfund Europe, one of the biggest intermediaries involved in securing funding for CBIJ in Europe, the news report "needs to be relevant for the European audience.It is the only limitation in the topic" (EUROPE Ltd & Media Consulting Group (Hepp & Loosen, 2021)ones able to invest the extra cost, time, and work required to secure funding from foundations through dedicated business teams while coordinating large, diverse groups operating in multiple countries.While the "local" phenomenon appeared as a buzzword in 2019, highlighting the added value of adapting global topics to the local context, most conference participants attending the Dataharvest sessions and workshops belong to networks that are highly institutionalised: International Consortium of Investigative Journalists, Organized Crime Corruption Reporting Project, Global Investigative Journalism Network, Investigate Europe, etc.This finding contradicts H3 of this study positing a higher level of autonomy for journalists thanks to initiatives outside stable and consolidated structures, as short-term partnerships, fluid work rhythms, non-hierarchical collaborations, and pioneering grassroots experiences(Hepp & Loosen, 2021)are displaced by "top media organizations"(Heft, 2021, p. 470).It is also important to note that this study has not identified any specific collaborative practices.The results only suggest that Dataharvest gives considerable attention to data collection, processing, and visualisation, as an essential part of the work of cross-border networks.Sessions aimed at training attendees to work with data (in its many forms) have been a constant at the successive conferences, with a total of 420 events offered, the largest numbers being in 2017(66), 2018 (56), and 2019 (50).This finding is similar to the information provided by Data Journalism and European Journalism Centre (2022) in its State of Data Journalism Survey, which highlights the skill deficits of journalists in this area and the need to address this problem.The incorporation of data into intertwined newsrooms has become indispensable because it facilitates a considerable increase in both the quantity and the quality of the stories produced