Transcript Madden - The Beckman Report on Database Research

Thoughts on the State of the DB
Community
Sam Madden
[email protected]
Thoughts on Community
• Conferences are too diffuse
– Single-track conferences provide a vastly superior experience
MIT COMPUTER SCIENCE AND ARTIFICIAL INTELLIGENCE LABORATORY
Thoughts on Community
• Conferences are too diffuse
– Single-track conferences provide a vastly superior experience
• Use OS community as an example
MIT COMPUTER SCIENCE AND ARTIFICIAL INTELLIGENCE LABORATORY
Thoughts on Community
• Conferences are too diffuse
– Single-track conferences provide a vastly superior experience
• Use OS community as an example
– Much higher impact per paper
* SOSP (56 cites per paper) is 2x SIGMOD (26 cites per paper
MIT COMPUTER SCIENCE AND ARTIFICIAL INTELLIGENCE LABORATORY
Thoughts on Community
• Conferences are too diffuse
– Single-track conferences provide a vastly superior experience
• Use OS community as an example
– Much higher impact per paper
* SOSP (56 cites per paper) is 2x SIGMOD (26 cites per paper
– Top authors have 1/3rd the pubs (Liskov = 13 SOSP+OSDI)
http://from-a-to-remzi.blogspot.com/2013/05/the-systems-top-50.html
MIT COMPUTER SCIENCE AND ARTIFICIAL INTELLIGENCE LABORATORY
Thoughts on Community
• Conferences are too diffuse
– Single-track conferences provide a vastly superior experience
• Use OS community as an example
– Much higher impact per paper
* SOSP (56 cites per paper) is 2x SIGMOD (26 cites per paper
– Top authors have 1/3rd the pubs (Liskov = 13 SOSP+OSDI)
http://from-a-to-remzi.blogspot.com/2013/05/the-systems-top-50.html
Proposal: create a “super conference”;
single track,
where it is an honor to present,
that every one wants to attend.
MIT COMPUTER SCIENCE AND ARTIFICIAL INTELLIGENCE LABORATORY
Thoughts on Community
• Conferences are too diffuse
– Single-track conferences provide a vastly superior experience
• Use OS community as an example
– Much higher impact per paper
* SOSP (56 cites per paper) is 2x SIGMOD (26 cites per paper
– Top authors have 1/3rd the pubs (Liskov = 13 SOSP+OSDI)
http://from-a-to-remzi.blogspot.com/2013/05/the-systems-top-50.html
Proposal: create a “super conference”;
single track,
where it is an honor to present,
that every one wants to attend.
Think CIDR but where the papers are actually good
And from the whole community
MIT COMPUTER SCIENCE AND ARTIFICIAL INTELLIGENCE LABORATORY
Thoughts on Research
• “Big Data” continues to lead CS
• By defining “databases” as “relational data management
systems”, we lose out
• We can do better at getting general data intensive
computing folks to participate in our community
• A “good” database paper shouldn’t have to be full of
psuedomath
• SOCC is on the right track, but cloud computing isn’t hip
anymore
MIT COMPUTER SCIENCE AND ARTIFICIAL INTELLIGENCE LABORATORY