TensorFlowSNA

A cross-disciplinary research project

About

TensorFlowSNA is a research project mining the TensorFlow software ecosystem with Social Network Analysis
Image of a neural network

Motivation

  • Many firms develop open-source platforms. But why and how big-tech giants like Google open-sourced advanced and complex technological platforms that started in-house? In other words, why releasing a complex and advanced technology with a open-source license and work with the community after so many internal investments?
  • Many rival firms simultaneously collaborate and compete in the open-source arena, i.e. open-coopetition (Teixeira, J. A. 2023). Why and how do rival technology vendors joint-develop the TensorFlow machine learning platform?

Key references

Key theoretical references
  • Teixeira, J. A. (2023). Towards understanding open­-coopetition -- Lessons from the automotive industry. in Proceedings of the 44th International Conference on Information Systems (ICIS 2023) AIS. Open-access right here.
  • Teixeira, J. A., Ahmed, S. S., Laine-Kronberg, L., Mezei, J., & Smailhodzic, E. (2025). Towards understanding open and coopetitive platform ecosystems: The case of TensorFlow. in Proceedings of the 33th European Conference on Information Systems (ECIS 2025) AIS. Open-access right here.
  • Jacobides, M. G., Cennamo, C., & Gawer, A. (2018). Towards a theory of ecosystems. Strategic Management Journal, 39(8), 2255-2276. Open-access at https://onlinelibrary.wiley.com/doi/full/10.1002/smj.2904.
Key methodological references
  • Teixeira, J., Robles, G., & González-Barahona, J. M. (2015). Lessons learned from applying social network analysis on an industrial Free/Libre/Open Source Software ecosystem. Journal of Internet Services and Applications, 6, 1-27. Open-access at https://link.springer.com/article/10.1186/s13174-015-0028-2.
  • Teixeira, J., Mian, S., & Hytti, U. (2016). Cooperation Among Competitors in the Open-Source Arena: The Case of Openstack. In Proceedings of the 37th International Conference on Information Systems (ICIS 2016) AIS. Open-access at https://arxiv.org/abs/1612.09462 .
  • Herbold, S., Amirfallah, A., Trautsch, F., & Grabowski, J. (2021). A systematic mapping study of developer social network research. Journal of Systems and Software, 171, 110802. Closed-access at https://www.sciencedirect.com/science/article/pii/S0164121220302077 .
  • Zhu, J., & Wei, J. (2019, May). An empirical study of multiple names and email addresses in oss version control repositories. In 2019 IEEE/ACM 16th International Conference on Mining Software Repositories (MSR) (pp. 409-420). IEEE. Closed-access at https://ieeexplore.ieee.org/abstract/document/8816766.
Related working papers from others
  • Osborne, C., Daneshyan, F., He, R., Ye, H., Zhang, Y., & Zhou, M. (2025). Characterising open source co-opetition in company-hosted open source software projects: the cases of PyTorch, TensorFlow, and transformers. Proceedings of the ACM on Human-Computer Interaction, 9(2), 1-30. Open-access at https://doi.org/10.1145/3710944.

Aim, objectives and research questions

Overall aim:
  • Advance our understanding on open-source, and open-coopetition by looking at the case of TensorFLow;
Objectives:
Primary objectives:
  • Contribute to open-source motivations literature;
  • Contribute to open-coopetition literature;
  • Document the history of TensorFLow;
  • Advance the mining of software repositories with Social Network Analysis;
Secondary objectives:
  • Contribute to platform ecosystems literature;
  • Contribute to coopetition literature;
  • Contribute to open-innovation literature;
  • Contribute to triple helix innovation literature;
  • Improve existing open-source tools that mine software repositories;
Research questions:
Regarding tech giants
  • Why do tech giants like Google open-source advanced and complex technological platforms that started in-house?
  • Why are different organizations cooperating with competitors in the co-production of advanced technological platforms?
Regarding non-commercial organizations
  • How did non-commercial entities contribute to the development of the TensorFlow open and coopetitive platform for AI and ML?
  • How do the different freedoms of open and coopetitive platforms align with the different interests of commercial, non-commercial and governmental organizations?

Research team

By chronological order as contributors worked on the project:

Methodological overview

Mix-methods approach

We combined and virtual-ethnography (VE) with a Social Network Analysis (SNA) over publicly-available and naturally-occurring open-source data that allowed us to re-construct and visualize the evolution of the TensorFlow collaboration as a sequence of networks. Knowledge from the VE informed the SNA and the other way around as we attempted not only to retrieve collaborative networks but also to interpret and explain them. We will also engage with active developers and a community manager to validate our preliminary results and findings.

Virtual Ethnography

We started by screening, by virtual ethnographic manners, publicly available data such as company announcements, financial reports and specialized-press that allowed us to gain insights of the industrial context. Then we could better design the mining of software repositories with SNA.

Social Network Analysis

After attaining a better understanding of the competitive dynamics of the Cloud Computing Industry, we started extracting and analysing the social network of the OpenStack community leveraging SNA (Scott, 2012; Wasserman and Faust, 1994), which is an emergent method widely established across disciplines of social sciences in general (Borgatti and Foster, 2003; Uzzi, 1996; Wasserman and Faust, 1994; Watts, 2004)

For understanding the evolution of the code-based collaboration, we connect developers who work on the same file, constructing a network of collaboration activities among developers. With the visualization of the network over time, we gain insights on collaboration and rivalry within the software project.

How we modeled the network
Modelling collaboration from the source-code repositories change-log

The collaborative network during a certain time slice can be formally defined as:
Gt = (V,Av,E)
Where:
V = A set of nodes representing the developers contributing to the TensorFlow open-source software project
E = A set of edges, identifying the connections between two developers if they have worked on the same software source-code file.
Av = A set of nodes-attributes, capturing each developer’s company affiliation. This information is extracted from the email address of each developer.

Semi-Structured interviews

Targeting software developers (i.e., code contributors to TensorFlow core) and program managers at the top 10 firms.

Guiding questionnaire for max 40 min interview:

  1. I assume from the data I have collected so far, that you are a contributor TensorFlow? Is that correct?
  2. What kind of contributions have you added to TensorFlow? Can you give examples?
  3. What motivated you to contribute?
  4. Are you the only one contributing to TensorFlow at your organization, or do you have a team that contributes regularly to TensorFlow?
  5. Will your contributions contribute to your career advancement?
  6. Would it be possible to file a patent related to your contribution?
  7. Are you worried that others might make money with your contribution?
  8. Do you perceive some inter-individual or inter-organizational competition in TensorFlow or is it all about collaboration?
  9. Do you consider contributing more to TensorFlow in the future? What about other projects?
  10. Do you think that governments should encourage contributions to open-source communities from universities? What about commercial companies?
  11. What are the main barriers that an AI or ML researcher faces when contributing to an open-source project?
  12. In what way do AI and ML researchers depend on the for-profit industry?
FINAL: I created the following additional social network visualizations and reports about collaboration in the TensorFlow ecosystem. Would you mind taking a fast look at them?

Real time transcription using Wisper

Tools

Tools for mining git repositories with SNA
Tools for the visualization of social networks
  • visone is a software for the visual creation, transformation, exploration, analysis and representation of network data, jointly developed at the University of Konstanz and the Karlsruhe Institute of Technology.
  • Tulip is an information visualization framework dedicated to the analysis and visualization of relational data. Tulip aims to provide the developer with a complete library, supporting the design of interactive information visualization applications for relational data that can be tailored to the problems he or she is addressing. Developed by LaBRI, University of Bordeaux, France.
Tools for the statistical analysis of social networks
  • statnet is a suite of open source R-based software packages for network analysis, along with a comprehensive set of training materials. Developed by Pavel Krivitsky, Skye Bender-deMoll, Michał Bojanowski, Carter T. Butts, Steven M. Goodreau, Mark S. Handcock, David R. Hunter, Chad Klumb, and Martina Morris among others.
  • Goldfish is a software tool (i.e. R package) for the analysis of time-​stamped network data using a variety of models. In particular, it implements different types of Dynamic Network Actor Models (DyNAMs), a class of models that is tailored to the study of actor-​oriented network processess through time. Goldfish also implements different versions of tie-​oriented relational event models. Developed by members of the Chair of Social Networks at ETH Zürich and James Hollway at the Graduate Institute in Geneva.

Results

TOP 10 organizational contributors that cooperate with others in the co-production of TensorFlow

First in terms of number of nodes/developers, then in terms of Lines of Code * Number of Commits:
  1. Google (cloud services, software tools, training and consulting) - 970 developers to TensorFlow core
  2. Intel (chipsets) - 93 developers to TensorFlow core
  3. Nvidia (chipsets) - 61 developers to TensorFlow core
  4. ARM (chipsets) - 48 developers to TensorFlow core
  5. IBM (cloud services, software tools, training and consulting)- 41 developers to TensorFlow core
  6. AMD (chipsets) - 27 developers to TensorFlow core
  7. Microsoft (cloud services, hardware, software tools, training and consulting) - 16 developers to TensorFlow core
  8. Huwaei (cloud services, hardware, software tools) - 12 developers to TensorFlow core
  9. Amazon (cloud services, software tools, training and consulting) - 11 developers to TensorFlow core
  10. Naver (cloud services, software tools) - 11 developers to TensorFlow core

TOP 10 non-commercial contributors that cooperated with others in the co-production of TensorFlow

First in terms of number of nodes/developers, then in terms of Lines of Code * Number of Commits:
  1. Seoul National University - SNU (Korean: 서울대학교)
  2. German Research Center for Artificial Intelligence (German: Deutsches Forschungszentrum für Künstliche Intelligenz - DFKI)
  3. Chromium - a free and open-source web browser project
  4. Institute for System Programming - ISP of the Russian Academy of Sciences (Russian: Институт системного программирования)
  5. Peking University - PKU (Simplified Chinese :北京大学)
  6. Georgia Institute of Technology - GeorgiaTech
  7. Massachusetts Institute of Technology - MIT
  8. Apache Software Foundation - an American nonprofit corporation to support a number of open-source software projects
  9. Royal Institute of Technology (Swedish: Kungliga Tekniska högskolan - KTH)
  10. University of California, Berkeley

Social network visualizations - Both inter-individual and inter-organizational collaborative networks

Inter-individual collaborative networks

The overall history of TensorFlow (7 Nov 2013 - 12 April 2024)
Captures collaboration between developers during the covered overall history of TensorFlow.
Circa 9 years of code-collaboration (7 Nov 2013 - 12 April 2024).
Circular layout. Nodes coloured by organizational affiliation.
Bots were filtered but developers names were not consolidated yet.
See that 554 developers are incorrectly affiliated with users!
Captures collaboration between developers during the covered overall history of TensorFlow.
Circa 9 years of code-collaboration (7 Nov 2013 - 12 April 2024).
Circular layout. Nodes coloured by organizational affiliation.
Bots were filtered and developers names were consolidated yet.
See Setting your commit email address documentation from GitHub Docs for an explanation on why so many commits were associated with USERNAME@users.noreply.github.com.
Captures collaboration between developers during the covered overall history of TensorFlow.
Circa 9 years of code-collaboration (7 Nov 2013 - 12 April 2024).
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
All dev. affiliated with all organizations.
Captures collaboration between developers during the covered overall history of TensorFlow.
Circa 9 years of code-collaboration (7 Nov 2013 - 12 April 2024).
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow.
All dev. affiliated with ['google','microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].
Captures collaboration between developers during the covered overall history of TensorFlow.
Circa 9 years of code-collaboration (7 Nov 2013 - 12 April 2024).
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow excluding Google.
All dev. affiliated with ['microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].

Last months of 2015

Captures collaboration between developers during last months of 2015.
Circular layout. Nodes coloured by organizational affiliation.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during last months of 2015.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during last months of 2015.
First commit Sat Nov 7 by Vijay Vasudevan (vrv AT google.com) starts TensorFlow as an open-source project.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow.
All dev. affiliated with ['google','microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].

2016

Captures collaboration between developers during last months of 2016.
Circular layout. Nodes coloured by organizational affiliation.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during last months of 2016.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during 2016.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow.
All dev. affiliated with ['google','microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].
Captures collaboration between developers during 2016.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow excluding Google.
All dev. affiliated with ['microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].

2017

Captures collaboration between developers during 2017.
Circular layout. Nodes coloured by organizational affiliation.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during of 2017.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during 2017.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow.
All dev. affiliated with ['google','microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].
Captures collaboration between developers during 2017.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow excluding Google.
All dev. affiliated with ['microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].

2018

Captures collaboration between developers during of 2018.
Circular layout. Nodes coloured by organizational affiliation.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during of 2018.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during 2018.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow.
All dev. affiliated with ['google','microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].
Captures collaboration between developers during 2018.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow excluding Google.
All dev. affiliated with ['microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].

2019

Captures collaboration between developers during of 2019.
Circular layout. Nodes coloured by organizational affiliation.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during of 2019.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during 2019.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow.
All dev. affiliated with ['google','microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].
Captures collaboration between developers during 2019.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow excluding Google.
All dev. affiliated with ['microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].

2020

Captures collaboration between developers during of 2020.
Circular layout. Nodes coloured by organizational affiliation.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during of 2020.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during 2020.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow.
All dev. affiliated with ['google','microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].
Captures collaboration between developers during 2020.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow excluding Google.
All dev. affiliated with ['microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].

2021

Captures collaboration between developers during of 2021.
Circular layout. Nodes coloured by organizational affiliation.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during of 2021.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during 2021.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow.
All dev. affiliated with ['google','microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].
Captures collaboration between developers during 2021.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow excluding Google.
All dev. affiliated with ['microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].

2022

Captures collaboration between developers during of 2022.
Circular layout. Nodes coloured by organizational affiliation.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during of 2022.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during 2022.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow.
All dev. affiliated with ['google','microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].
Captures collaboration between developers during 2022.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow excluding Google.
All dev. affiliated with ['microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].

2023

Captures collaboration between developers during of 2023.
Circular layout. Nodes coloured by organizational affiliation.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during of 2023.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during 2023.
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow.
All dev. affiliated with ['google','microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].
Captures collaboration between developers during 2023. Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow excluding Google.
All dev. affiliated with ['microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].

Last months 2024

Captures collaboration between developers during first months of 2024 (Jan-Apr).
Circular layout. Nodes coloured by organizational affiliation.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during first months of 2024 (Jan-Apr).
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered but developers names were not consolidated yet.
Captures collaboration between developers during first months of 2024 (Jan-Apr).
Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow.
All dev. affiliated with ['google','microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].
Captures collaboration between developers during first months of 2024 (Jan-Apr). Centrality layout. Nodes coloured by organizational affiliation and sized by centrality.
Nodes position by Fruchterman-Reingold force-directed algorithm.
Bots were filtered and developers names were consolidated.
Limited to the top 10 contributors that compete with others in the co-production of TensorFLow excluding Google.
All dev. affiliated with ['microsoft','ibm','amazon','intel','amd','nvidia','arm','meta','bytedance'].

Inter-organizational collaborative networks

Exploring the collaborative networks of non-commercial organizations (7 Nov 2013 - 12 April 2024)
  • Seoul National University - SNU (Korean: 서울대학교)
    • snu
    snu.png - Nodes are organizations, edges are collaborative relationships, edges-weights are unique pairs of developers into a dyadic inter-organizational relationship (i.e., intensity of collaboration).
  • German Research Center for Artificial Intelligence (German: Deutsches Forschungszentrum für Künstliche Intelligenz - DFKI)
    dfki
    dfki.png - Nodes are organizations, edges are collaborative relationships, edges-weights are unique pairs of developers into a dyadic inter-organizational relationship (i.e., intensity of collaboration).
  • Chromium - a free and open-source web browser project
    chromium
    chromium.png - Nodes are organizations, edges are collaborative relationships, edges-weights are unique pairs of developers into a dyadic inter-organizational relationship (i.e., intensity of collaboration).
  • Institute for System Programming - ISP of the Russian Academy of Sciences (Russian: Институт системного программирования)
    ispras
    ispras.png - Nodes are organizations, edges are collaborative relationships, edges-weights are unique pairs of developers into a dyadic inter-organizational relationship (i.e., intensity of collaboration).
  • Peking University - PKU (Simplified Chinese :北京大学)
    pku
    pku.png - Nodes are organizations, edges are collaborative relationships, edges-weights are unique pairs of developers into a dyadic inter-organizational relationship (i.e., intensity of collaboration).
  • Georgia Institute of Technology - GeorgiaTech
    gatech
    gatech.png - Nodes are organizations, edges are collaborative relationships, edges-weights are unique pairs of developers into a dyadic inter-organizational relationship (i.e., intensity of collaboration).
  • Massachusetts Institute of Technology - MIT
    mit
    mit.png - Nodes are organizations, edges are collaborative relationships, edges-weights are unique pairs of developers into a dyadic inter-organizational relationship (i.e., intensity of collaboration).
  • Apache Software Foundation - an American nonprofit corporation to support a number of open-source software projects
    apache
    apache.png - Nodes are organizations, edges are collaborative relationships, edges-weights are unique pairs of developers into a dyadic inter-organizational relationship (i.e., intensity of collaboration).
  • Royal Institute of Technology (Swedish: Kungliga Tekniska högskolan - KTH)
    kth
    kth.png - Nodes are organizations, edges are collaborative relationships, edges-weights are unique pairs of developers into a dyadic inter-organizational relationship (i.e., intensity of collaboration).
  • University of California, Berkeley
    berkeley
    berkeley.png - Nodes are organizations, edges are collaborative relationships, edges-weights are unique pairs of developers into a dyadic inter-organizational relationship (i.e., intensity of collaboration).

Expected contributions

Expected theoretical contributions

Preliminary answers to the guiding research questions:
  • Why do tech giants like Google open-source advanced and complex technological platforms that started in-house?
    • Extended R&D reach;
    • Extending the size of the market;
    • Creating demand for complementary products and services (e.g., computing services, chip-design, quality labelled data, AI/ML models);
    • Finding external complementarities;
    • Providing strong arguments for future anti-trust cases;
    • Easier cooperation and integration with academia;
    • Extended reputation in interactions with academia;
    • Easier talent identification and evaluation;
  • Why are different organizations cooperating with competitors in the co-production of advanced technological platforms?
    • In the case of competing chip-makers, like Nvidia, Intel, ARM and AMD, not co-cooperating would mean that TensorFlow loads would run in the chips of competitors.; It is of all chip-makers interest to insure that TensorFLow runs on their chips;
    • The same for vendors of hardware enabling AI/ML (e.g., servers or specialized boards);
    • The same for vendors of hardware with AI/ML features (e.g., cars and tools with computer vision recognition);
    • The same for vendors of hosted computing services;
    • Also, by cooperating with competitors, in certain conditions, the size of the market also extends via extended networks reach;
    • Avoid the fragmentation of AI/ML frameworks. Standarization

Expected methodological contributions

  • The use of Fruchterman-Reingold force-directed algorithm allow the identification of small isolated sub-communities in the TensorFlow community on the fly. Something that was is not captured on visualizations based on degree centrality as in Teixeira et al. (2015) and Teixeira et al.(2016).
  • A simple but yet elegant way of combining quantitative, quantitative and relational social network data.

Publications

By chronological order as results got published:

  • Teixeira, J. A., Ahmed, S. S., Laine-Kronberg, L., Mezei, J., & Smailhodzic, E. (2025). Towards understanding open and coopetitive platform ecosystems: The case of TensorFlow. in Proceedings of the 33th European Conference on Information Systems (ECIS 2025) AIS (conditionally accepted 28 Feb 2025). Open-access right here.

Links

Related master thesis projects
Wikipedia pages on open-source and open-coopetition
Past related projects
Blogs and articles on the history of TensorFlow
Blogs on the TensorFlow way of working
Blogs on why TensorFlow is open-source
Blogs on why TensorFlow is coopetitive
  • Any?
Videos on what is TensorFlow
Videos on the history of TensorFlow
Videos on why TensorFlow is open-source
Videos on TensorFlow way of working
Tutorials for master students and doctoral students joining the project
Videos on theory for master students and doctoral students joining the project
On the triple-helix innovation model
On open-innovation
On open source software
On coopetition
On open-coopetition
Websites on theory for master students and doctoral students joining the project

Contact

Jose Teixeira < jose.apolinarioteixeira AT aalto.fi >