Industrial Labs to Academia Academia to Industrial Labs

1
Industrial Labs to Academia?Academia to
Industrial Labs?

• A Study of the Movement of Database Researchers

William Conner
2
Question

• Statement Usually, its harder to move from
  industrial research labs back to academia rather
  than the other way around
• Made by UIUC faculty member
• PhD Job Hunt Seminar on October 6, 2003
• Does this statement hold for the database
  community?

3
Methodology

• Survey the database staff at the top universities
  and top industrial labs
• 19 universities, 4 industrial labs
• Looked at CVs and bios
• Grad student affiliation and post-docs were not
  counted
• Counted transitions in both directions

4
Methodology (contd)

• Finding top universities
• Top-25 overall CS schools with DB groups
• U.S. News and World Report Rankings
• 6 didnt have links to DB groups on homepage
• Four well-known labs in DB
• Encountered additional places during this search
  by chance
• e.g., NEC Labs, Santa Clara University

5
University DB Groups

• Berkeley
• Brown
• Carnegie Mellon
• Columbia
• Cornell
• Duke
• Georgia Tech
• Illinois
• Maryland

• Massachusetts
• Michigan
• North Carolina
• Penn
• Purdue
• Stanford
• UCLA
• UC- San Diego
• Washington
• Wisconsin

6
Industrial Lab DB Groups

• ATT Labs
• Bell Labs
• IBM
• Microsoft

7
Disclaimers

• Many researchers dont make their biographical
  info readily available
• Found them, but excluded them from investigation
• Hard to navigate through industrial labs
• Broken links for DB groups and staff
• Had to find people on SIGMOD organization
  committees and use search engines
• Many more excluded from investigation because I
  couldnt find them at all

8
Results

• Found 19 database researchers that have moved
  between academia and industrial labs
• Visited over 70 homepages of DB people
• Some moved more than once
• Industrial Labs to Academia
• 16 transitions
• Academia to Industrial Labs
• 6 transitions
• Based on this ad-hoc study, industrial labs to
  academia moves are more common than other
  direction

9
Results (contd)

• Most database researchers stay on one side of the
  aisle
• Many did not move at all
• Started career at university or industrial lab
  and stayed put
• Many stayed in academia
• University to university
• Lab to lab moves rare
• Unless you want to count ATT vs. Lucent

10
Results (contd)

• Largest outflows
• Bell Labs had 8 researchers leave
• A share of Lucent stock was less than a dollar
  during part of October 2002
• 5 left Bell Labs between 2002 and 2003
• IBM had 4 researchers leave
• Includes both T.J. Watson and Almaden
• No major inflows

11
Results (contd)

• UIUC faculty member who made statement
• NEC Research Institute -gt UIUC
• DB researcher from Princeton
• Most transitions in this study
• IBM -gt Texas -gt Bell Labs -gt Lehigh

12
Data

• Ted Johnson (Florida -gt ATT Labs)
• Avi Silberschatz (Texas -gt Bell Labs -gt Yale)
• Hank Korth (IBM T.J. Watson -gt Texas -gt Bell Labs
  -gt Lehigh)
• Richard Hull (USC -gt Bell Labs)
• Juliana Freire (Bell Labs -gt OGI)
• Gautam Das (U. of Memphis -gt Microsoft)
• Jayavel Shanmugasandaram (IBM Almaden -gt Cornell)
• Alan Demers (Xerox PARC -gt Cornell)
• Peter Haas (Santa Clara -gt IBM Almaden)

13
Data (contd)

• Jeffrey Ullman (Bell Labs -gt Princeton -gt
  Stanford)
• Jennifer Widom (IBM Almaden -gt Stanford)
• Yuanyuan Zhou (NEC Research Institute -gt
  Illinois)
• Wesley Chu (Bell Labs -gt UCLA)
• Carlo Zaniolo (Bell Labs -gt MCC -gt UCLA)
• Alon Halevy (ATT Labs -gt Washington)
• H.V. Jagadish (ATT Labs -gt Illinois -gt Michigan)
• Wei Wang (IBM T.J. Watson -gt North Carolina)
• Chee-Yong Chan (Bell Labs -gt National Univ. of
  Singapore)
• Prasan Roy (Bell Labs -gt IIT-Mumbai)