Data-driven Networking Research: models for academic collaboration with industry

Metadata

Title: Data-driven Networking Research: models for academic collaboration with industry (a Google point of view)

Presented in SIGCOMM Computer Communication Review 2021.

Authors: Jeffrey C. Mogul, Priya Mahadevan, Christophe Diot, John Wilkes, Phillipa Gill, Amin Vahdat (Google)

Understanding the paper

The biggest help academia can get from industry is their scenarios and real-world, large-scale data.

Google explains several constraints when sharing data with the academia:

• Hard constraints

  1. Privacy (even the interns and visiting scientists never have access to user data)

  2. Business concerns (even anonymizing the data still has risks to expose technical information or business-growth information)

• Soft constraints

  1. Scale (the collection of large-scale infrastructural data is expensive to build and operate)

  2. Operational Risk (even seemingly-safe changes may cause problems)

  3. Staff time (releasing high-quality datasets will cost much time)

All in all, they think working temporarily as employees such as interns and visiting scientists will be the best option. While here they focus on the field of network research, I believe it is a common problem if we're going to do some collaborations between the academia and industry.

Maybe if we want to do some real-world research in academia, just work as a temporal employee with little money and directly handle these dirty data in the industry?