The role of industry and academia in network research
Title : The role of industry and academia in network research
Host : Zili Meng (Hong Kong University of Science and Technology), Xiaochong Jiang (Zhejiang University)
Panelists : Ammar Tahir (University of Illinois Urbana-Champaign), James Larisch (Harvard University), Ye Yang (Alibaba Cloud), Xiaochong Jiang (Zhejiang University), Zahaib Akhtar (Amazon Prime Video)
Scribe: Rulan Yang, Gao Han (Xiamen University)
Introduction
This session is about how the collaboration between academics and industry works and to identify if there is any chance that we can exploit or utilize it later.
The conference data show that the ratio of papers from collaborations between academia and industry has significantly increased. Therefore, this session addresses balancing academic research with industrial applications and how collaboration can drive technological innovation. Four authors of Sigcomm’24 were invited to share their collaboration experiences, including how their partnerships began. Their insights are intended to help us better understand how to collaborate effectively.
Ammar Tahir: Collaboration between the University of Illinois and Microsoft Research
Ammar shared his story of working together on the BC-PQP project. He noted that many of the problems are becoming more apparent at the industrial scale, and the solutions are becoming more feasible and effective at the scale at which the industry operates.
James Larisch: Collaboration between the Harvard University and Cloudflare
James shared his experience of interning and publishing at Cloudflare during his PhD. He believes that internships are important to help establish connections with industry, learn about real-world problems, and be able to use these problems for research during the PhD. In addition, internships provide experience that can be very useful for future career choices.
Ye Yang, Xiaochong Jiang: Collaboration between Zhejiang University and Alibaba
Ye supported the idea that collaboration can lead to experience on both sides of the fence. For example, knowing what kinds of problems are appropriate for research, what is a research problem and what is an engineer’s problem. And, knowing where the real gap is between a paper and actual deployment.
Xiaochong shared what he learned from the experience. Many of the theories or knowledge learned in school may only be implemented in a simulator. But in industry, he can see how the code works. Analyzing the problems encountered in the industry can identify a potential bottleneck. Finally, we can turn it into a new idea.
Zahaib Akhtar: Collaboration between the University of Massachusetts Amherst and Amazon Prime Video
Zahaib noted that companies also encourage collaboration with universities. Amazon collaborates with several universities and creates internships so that academic research and practical applications are closely integrated.
Questions and opinions
The first question is about how to find more practical problems. One of the panelists from industry gives a straightforward approach, that is to examine all real-world incidents to identify common patterns.
The second question is to discuss whether papers from industry are more likely to be accepted (asked by the host). And two panelists give “yes” answers certainly, especially for systems-focused papers. They explained that it’s easier to motivate a paper from industry because you can state, “We are a big company, and we had this problem and solved it.” rather than trying to convince reviewers that a problem exists and is worth solving. This eliminates any doubt for reviewers regarding the practical relevance of the problem.
The third question is literally about the role of industry and academia, the questioner asked “Will the industry lead the trend? In extreme cases, all papers are completed by industry collaborators. Does this mean that industry ultimately determines what needs to be created and how it should be done?”. Then the speakers acknowledge that this phenomenon does indeed exist to some extent, akin to the streetlight effect. This is because industry is driven by its own needs, not necessarily by research considerations.
Personal thoughts
As a student currently doing an internship in industry, I think it is a valuable opportunity to intern in a research department in an enterprise. This is because the problems in industry are far more complex than we had imagined. In real-life network and system usage, we encounter many corner cases that, although hard to solve, undoubtedly broaden our comprehensive understanding, especially when facing new scenarios and challenges. Besides, the industry benefits from quick feedback from a large number of users, which greatly aids research and problem-solving.
However, since problem-solving is a top priority in the industry, enterprises may sacrifice generality for efficiency to implement customized solutions toward specific user scenarios. This is where academic researchers need to step in to think about and improve the foundational aspects of the models accordingly.