Spotlight Series Recap: Open Source, Tech Transfer & Commercialization
On November 15th, 2022, the Higher Education Leadership Initiative for Open Scholarship (HELIOS Open) convened academic leaders to discuss their open source, code, and software sharing efforts. This blog post summarizes key themes from the session.
The session was moderated by Drew Endy, Martin Family University Fellow in Undergraduate Education at Stanford University, where he is also faculty co-director of degree programs at the Hasso Plattner Institute of Design (aka Stanford’s d.school). He’s also served as a member of NASEM’s Standing Committee on Science, Technology, and Law. As a bioengineer working in synthetic biology, Dr. Endy and his teams have made many contributions to open biotechnology, especially with material transfer agreements (MTA), which are typically bilateral contracts that govern the sharing of physical samples used in life science research. Most MTAs in biotech and academia prohibit redistribution of received materials and also sharing with commercial partners. While often appropriate, these restrictions can hinder translation of materials that could otherwise be freely shared.
Thus, Drew’s team created the OpenMTA as an option to conventional MTAs. The OpenMTA purposefully allows both redistribution of received materials and distribution to commercial partners. About 100 institutions already support the OpenMTA so that students and researchers have the option of more readily sharing materials in support of innovation and translation, which are shared goals of our collective HELIOS Open work.
The next contributor was Alexander Amini, Postdoctoral researcher at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL)
Dr. Amini described creating and open sourcing VISTA to the public. He and his team are working on and have open sourced a data-driven simulator, built and rendered using raw data from the real world. Their vision for VISTA is that it serves as a step toward building a more sustainable and resilient mobility path for the future of transportation.
With autonomous vehicles and robotics, realizing a society with embedded autonomous vehicles can be difficult. You can train your systems in the real world, but doing so is both time intensive and dangerous. Simulation presents opportunities to train in a much safer, controlled environment. Yet, reality gaps exist even today's best simulators, discouraging direct transfer into the real world. The VISTA approach leverages real data sets to build and scalably engineer synthetic simulated worlds.
Developing this fully data-driven approach presents a very interesting problem for open source, Amini described, because you're not just open sourcing software; you are also open sourcing the data that drives that software. VISTA can harness the power of the data available to synthesize these highly realistic, high fidelity, and scalable data sets, including data on cases that are normally too expensive and dangerous to collect when testing in the real world.
In the autonomous vehicle community, autonomous vehicle companies, and those that are considered leading pioneers, are building up closed source simulation engines. Without unifying across these engines, and without a unified testing framework for different pipelines, it is difficult for the government to regulate and evaluate what it means to have a good autonomous solution that can be deployed into society. By open sourcing the code, Amini and team are enabling safe and data-driven environments, allowing creators to unify around good governance of artificial intelligence policies.
Since open sourcing VISTA a few months ago, it has amassed over 7,000 independent installations with about a 100 new installations of the simulation engine every week. The team is excited about VISTA’s unique ability to create immersive virtual worlds.
Following Dr. Amini was Julieta Arancio, Postdoctoral researcher at Drexel University’s Center for Science, Technology and Society and at the University of Bath
Open hardware is the practice of licensing the designs of a physical object in a way that allows the object to be studied, modified, created, and distributed by anyone.
Dr. Arancio began by reflecting on an academic culture where commercialization is embedded into research training early and by design. When researchers develop a new design that includes a hardware component, they work with Technology Transfer Offices (TTO), and the TTO decides if the invention is patentable or not. It is a huge investment for colleges and universities, and the process has several challenges, including a lack of transparency and other considerations:
Journals often do not include ancillary or contextual information about hardware designs, which is needed for reproducibility.
The current academic culture can encourage secrecy, which slows down innovation, and rewards commercialization over sharing and collaboration. Some key designs are protected, precluding creation on top of the original design.
Choosing not to patent some inventions leads to missed opportunities for greater impact.
In advocating for open hardware, Arancio encouraged researchers to think beyond, “is my invention patentable or non patentable?” and look to opportunities. Consider openly licensing hardware designs with TTOs as partners. Good open hardware design is accompanied by documentation, including design files, the source code, and other layers of instruction. This lowers the barrier for collaboration and allows new communities to emerge. Open hardware also encourages reproducibility and innovation.
In the last 5 years we have seen open science hardware business models emerge. Arancio provided three examples of open hardware businesses. These organizations operate by protecting trademarks and selling specific components (kits, devices, and other physical objects), design services, and technical support expertise.
Open Ephys, a company building open source tools for neuroscience. Hardware researchers are tinkering all the time with their tools, adapting them to new research questions. Open Ephys is showcasing and making accessible incredible designs that they think deserve more recognition. They also advocate for open standards. They provide technical support, training, and warranty.
OpenTrons, a company that produces robots for experimentation. You can add modules to your robot as needed. OpenTrons offers a common platform to easily share protocols and reproduce results. They are also advocate for open protocols that increase reducibility.
OpenFlexure Microscope, a company that offers a design for a fully 3D printed, customizable, open-source optical microscope. The microscope includes excellent documentation online and has a robust community supporting it. Users include community scientists in Argentina, medical doctors in the US, and malaria researchers in Tanzania.
Arancio asserted the importance of shifting the way we train and support researchers. We must show students that there is more to innovation than patenting, and that you can grow professionally if you do open work.
She closed with several thought-provoking questions:
Could we ask researchers to share designs through institutional repositories, could we ask funders to mandate open hardware, and can TTOs adopt open licensing practices? There are procurement strategies that can change to promote open hardware. Better science enables more research questions and the ability to access and modify existing tools to create new knowledge.
Our final speaker was Sayeed Choudhury, Director of the Open Source Program Office (OSPO) at Carnegie Mellon University
Dr. Choudhury spoke about Open Source Program Offices (OSPO), their relationship to TTOs, and opportunities for collaboration for impact. At a college or university, researchers work with TTOs to commercialize and grow impact. Choudhury argued that OSPOs can serve a major role, and asserted that software is the most important output to share to enable reproducibility.
Choudhury provided background on three new federal memos that include code and software sharing components:
The Department of Defense (DoD) has a long history of working with open source software. DoD affirms, from both the consumption and production perspective of software, that we need to be open by default. This kind of design principle was an important underpinning for the development of the Internet and the World Wide Web, with DoD involved in both.
The recent White House Office of Management and Budget (OMB) memo focuses on procurement issues for open source software. It includes a statement about “vendors” needing to attest that they are complying with the National Institute of Standards and Technology (NIST) framework for secure software. It is unclear who or what constitutes a “vendor” or how such attestations would be made for open source software.
The Nelson Memo from the White House Office of Science and Technology Policy (OSTP) builds upon the 2013 memo, and requires all federally funded research articles be open immediately upon publication in an agency-designated repository. The memo does not directly mention software, but does open up the option to share other research outputs like code and software. It represents public access interest shifting from articles to data. Software is arguably next in terms of attention and policy.
He believes we have a collective opportunity to influence the federal government, including the OSTP, as they move forward in their thinking about open source software. OSPO++, funded by the Sloan Foundation, is a network formed around university-based OSPOs and some government-based open source offices. He believes forming a network of universities to explore collective conversations around open source is a good approach to influencing and encouraging software sharing policies.
Choudhury put forward the following takeaways he learned from his own experience and conversations with the federal government:
We should be proactive, not reactive, in encouraging OSTP to develop software sharing policies. We need to work across sectors and among institutions in partnership to inform the next instantiation around open source software, particularly as it relates to lessons learned from responses for public access to articles and data.
Consider OSPOs as organizational APIs that can help design and support the OSTP’s consideration of public access to open source software, including agency designated repositories. We, as institutions, must position ourselves in this context and recognize that by doing so, and considering the private and corporate repositories that already exist, there are opportunities and also threats.
We must figure out the intellectual property issues around open software. We can learn from the private sector and organizations like the Open Source Initiative. Reflecting on institutions and their handling of data and data sharing requirements, he sees institutions trying to figure out how to open their data and believes we can do better, particularly as it relates to coordinated responses across universities.
Finally, Choudhury reflected on the balance between and among academic freedom, reproducibility, open scholarship, risk management, and technology transfer. We need to understand that university administration, TTOs, research projects administrators, and others think about risks in sharing research outputs. Choudhury envisions an institutional Venn diagram where one looks at risk alongside open scholarship benefits and academic freedom to find a sweet spot for action. Different institutions might have specific contexts or requirements (e.g., public or private university), but as a community, we can come together and work proactively to inform the OSTP, e.g., about how we wish to share software.
Choudhury also noted the difference between technology transfer, knowledge transfer, and innovation. Technology transfer typically involves commercialization that benefits an individual university. Knowledge transfer refers to translation or dissemination of open scholarship without commercialization, typically associated with the social sciences and humanities. Innovation refers to translation or dissemination that features new forms of social impact and partnerships (e.g., community centers) and national or global impact (e.g., university outputs licensed to help the US government rebuild manufacturing capacity).
Choudhury concluded with two invitations: 1) an invitation to join an OSPO++ webinar to learn more about institutional OSPOs, and 2) an invitation to join a working group focused on the big questions around open source software and knowledge transfer.
Please contact Caitlin@orfg.org for more information on both invitations, as well as any other questions or thoughts you may have related to open source and commercialization.
