Saturday, December 27, 2008

Research skills

Posted on Saturday Oct 04, 2008

Yesterday I gave a talk on research skills, whose slides are here. Feedback is welcome.


  1. Dear Tao,

    I am a student that is pursuing a Ph.D. degree in computer science. I am in the area of software testing and verification. In the past few years, I read hundreds of papers from ICSE, PLDI, POPL, FSE, CAV, etc. One question on the usefulness of research always confused me. I felt that most of papers are not practically at all -- the techniques that are proposed in the papers are either rather limited such that they cannot be applied into practice, or being good in theory but will not be success as a product in industry due to other factors such as the cost-effective. What do you think about this?

    Many thanks!

    Best regards,
    Andrew Bernstan

  2. Andrew,

    Most PhD students or even researchers do think this way at some point or the other. The following is my take on the interesting point you have raised:

    How many of the PL features, test automation techniques, SE practices, etc. you see today appeared first in these top tier conferences? Almost, all of them! Most such papers, which introduce breakthrough concepts are candidates for "most influential paper" awards. What leads to such breakthrough ideas? These award winning papers are actually based on previous work in the area, all of which might not be industrially feasible. So, not all ICSE/PLDI/OOPSLA... papers transform into industrially feasible tools and practices. But, however incremental, they do their best in advancing the frontiers of Computer Science as we know. They ultimately contribute to the techniques which make their way into everyday practice.


  3. Dr. Xie,

    I enjoyed your slides, and many of your posts in this blog. Thanks for sharing with us readers.

    a PhD student from U. of Michigan.