[Published originally in the November 2002 edition of Computing Research News, Vol. 14/No. 5, pp. 1, 9.]
The Intel Research Network: An Innovative Model of Industry-University Collaboration
By Hans Mulder
This is another in a series of CRN articles describing the activities of CRA's industry laboratory members. Others are posted at: http://www.cra.org/reports/labs.
Intel has a long history of funding academic research through sponsored programs and grants. Today more than 250 Intel-sponsored research engagements are underway at universities throughout the world. Recently we have developed a bold new approach to conducting joint research with universities in an open collaborative environment. We believe this innovative model will accelerate Intel's exploratory research efforts, while addressing the most pressing concerns of all parties involved in collaborative initiatives between companies and universities.
Formation of Intel Research
In 1999, David Tennenhouse joined Intel and was charged with launching a new internal organization, Intel Research, to explore the disruptive and emerging technologies that could advance Intel's business and create new markets and opportunities. Tennenhouse had been chief scientist and director of the information technology office at the Defense Advanced Research Projects Agency (DARPA), an organization known for its ability to catalyze innovation by funding highly targeted university research projects.
The Vision of Proactive Computing
Tennenhouse arrived at Intel with a vision of a future world of proactive computing, in which billions of devices embedded throughout the environment will anticipate people's needs and take action on their behalf. Intel Research set out to translate that vision into reality.
The research model developed for this long-term and exploratory program uses Intel funding to sample the broad array of university research. However, just sponsoring university research would not be sufficient to bridge the gap between the world of university ideas and Intel's R&D. To succeed, we would need to create an environment in which we could focus on some the most promising technologies, then work on them together with universities. In addition, we would need to create a research environment within Intel that could move some of these collaborative projects downstream, towards products.
Launching a Network of University Labs
After a year of exploring possibilities, we launched a new model of industry-university collaboration in the form of the Intel Research Network of university labs. These project-focused labs, wholly owned and funded by Intel, are located near major universities. The universities were selected for their expertise in specific areas of computer science and information technology research that support our proactive computing research agenda, and for their willingness to experiment with an open collaborative model of joint research.
Currently the network consists of four labs located adjacent to UC Berkeley, the University of Washington, Carnegie Mellon University, and Cambridge University. Each lab explores a different aspect of proactive computing, from new technologies for ubiquitous computing environments to software for widely distributed storage systems.
The labs' directors are top academic faculty with tenured appointments in their respective departments. During their tenure as Intel lab directors, they are Intel employees, on partial leave from their faculty positions. Each lab will have approximately 20 Intel researchers and an equal number of university researchers when fully built out. They will collaborate closely in an environment that promotes sharing of knowledge, a commitment to timely publication of results, and broad diffusion of research results. The research agenda of each lab will evolve to take advantage of new research opportunities. Since most of the lab directors are on two- to three-year leaves, the periodic rotation of directors will keep the agenda fresh.
To facilitate technology transfer from these laboratories, we have set up a bi-directional tech transfer structure that can either take lab technology downstream or infuse Intel technology into these labs and the universities.
Overcoming the IP Obstacle
The sticking point in all university-industry research relationships is intellectual property rights. The history of university-industry collaboration is rife with cases of procrastination on contracts interfering with useful research and/or resulting in surprising and problematic disputes. In some cases, no collaboration happens at all.
For the purpose of the network of laboratories, the four principles we have established are that: 1) collaboration--not just throwing money over the wall--should be the norm; 2) the IP of the research we collaborate on should be non-exclusive; 3) there are many areas of research where, for the sake of collaboration, we favor not filing for IP protection at all; and 4) proprietary advantage should be generated when the principles take ideas downstream--for example, through internal R&D programs.
Not all universities, licensing offices, and/or professors will be able to agree to these principles. Some professors like to keep their IP exclusive so they can use it to start their own companies. Some licensing offices strongly believe they can extract value from CS and EE patents. However, we have found a great number of researchers who share our principles and are very keen on collaboration.
At the CRA Snowbird conference in July, the popular argument that computer science IP generates substantial revenue for universities was deconstructed by J Strother Moore of the University of Texas at Austin. Moore analyzed the licensing income of universities and concluded that, while IP is indeed a powerful revenue generator for universities, only a fraction of the fees come from licensing computer science or electrical and computer engineering innovations. A similar analysis done by Dave Hodges, former Dean at UC Berkeley, led to significant changes in the flexibility of UC campuses in negotiating sponsors' rights to university intellectual property developed in CSEE-based sponsored research. (http://patron.ucop.edu/ottmemos/docs/ott00-02.html)
The vast majority of licensing income is generated by other fields--most prominently, biotechnology, agriculture, and medical technology--in which each product embodies a single, clearly defined, patentable invention. By contrast, CS/ECE products may rely on hundreds or thousands of ideas in various configurations. This makes it difficult to extract royalties, since each patent is a minor contributor and it is often possible to substitute alternative technologies. Furthermore, attempting to do so alienates companies that would sponsor university research if it weren't for all the restrictions. In fact, some companies have decreased or eliminated their research sponsorships for this reason.
With so little for universities to gain from restrictive agreements in CS/ECE and so much to lose in terms of industry sponsorship, why not collaborate openly? Leadership in the CSEE Departments at UC Berkeley and the University of Washington reached this same conclusion a few years ago. UC Berkeley incorporated it in their CITRIS plans and the University of Washington started the process with the Portolano Technology Access Program. That, in addition to their expertise, made them a natural partner to help launch our new network of university labs. Together we formulated an Open Collaborative Research (OCR) Agreement. That document serves as a template that has been tailored for each of the labs in the network. Intel has recently signed an open collaborative research agreement with UC Berkeley, University of Washington, and Carnegie Mellon University.
The OCR Agreement
The OCR agreement provides a framework for Intel and the university to conduct joint research in the open. Under the agreement, Intel and principal investigators within the university can propose and initiate computer science and information technology research projects they wish to jointly undertake. Each project is defined in a project document that specifies the boundaries of the project--specifically, what research is to be conducted in the open--with the expectation that the results will be published and made widely available. The OCR provides the master agreement that addresses how these projects are initiated, documented, and approved, including the identification of specific facilities and researchers involved. Our intention is to make it fast and easy for researchers on both sides to collaborate and move freely between Intel and campus to conduct joint research projects.
The emphasis of the agreement is on acceleration of research, bringing together the combined strengths of both Intel and the university, and on the timely publication of research results. Though patents are expected to be rare, as the focus of the collaborations is on openness, the document does spell out the process for handling patents when they arise. As noted earlier, the focus is on non-exclusive access to IP, both for participants and for parties outside the agreement.
The OCR agreement further encourages collaboration by allowing for third parties to participate (we're currently talking with one company that wants to participate in a project within one of our labs). It also can accommodate larger collaborations that form around major projects. For example, UC Berkeley's CITRIS program may consider using our agreement as a guide for managing the IP issues surrounding that project. In short, the OCR agreement is designed to make collaborating easy.
As the foregoing suggests, our open collaborative model provides benefits for all participants. It allows university researchers to amplify their thinking and their work--and potentially see it translated into commercial products--without having to leave academia. It enables Intel to accelerate research in areas we find interesting and worthy of exploration by conducting research concurrently in the labs and within our company. By facilitating synergy and open exchange of ideas, the model will enable Intel and the participating universities to jointly lead the industry, to generate breakthroughs that will continue to advance the state of the art. Under this new model of industry-university research, we believe everyone wins. (It is important to note, though, that we apply the model to our long-term engagements in exploratory CS/IT research. It is an open question whether or not our principles and this collaboration model can be applied to other engagements.)
Other companies and industry groups have expressed interest in our open approach, and in developing standard agreements governing industry-university collaboration. We believe this would be beneficial, both for our industry and for universities. We encourage others to build on the foundation we have laid, and to use our agreement as a starting point for building a set of standard documents covering a variety of research relationships.
Hans Mulder, Ph.D., is a Sector Director within Intel Research, responsible for driving research into ubiquitous computing and distributed systems. He also is co-director of Intel Research Berkeley, one of four labs currently in the Intel Research Network. For more information about the labs and their people and projects, visit www.intel-research.net.
Copyright © 2007 Computing Research Association. All Rights Reserved. Questions? E-mail: firstname.lastname@example.org.