1. How to select a research topic
Mythili Vutukuru
IIT Bombay

2. Interesting vs. important
A PhD topic should ideally be both
Interesting, intellectually challenging, and
Important, impactful, making the world better!
Hard to achieve both, so which one to choose?
No publications if not enough intellectual depth
A reasonable compromise: at least the problem should be important (the solution may be futuristic)
Make sure you are solving a REAL problem

3. “Real” problems
For a good thesis in a systems area, make sure your problem is real
Observable (by you or others) now or in near future
It is a big enough problem worth fixing
No obvious solutions exist
But not being solved by industry right away
But something they would like to solve in 5-10 years
Academia feeds ideas to industry

4. How to find relevant areas
Look for problems in new, upcoming, “hot” areas
Papers in top-tier conferences
Papers published by top research groups
How to judge if you are in the right area?
Can you get published in top-tier venues?
Can you get funded from top industry grants?
Avoid getting lost in your niche area that no one cares about

5. Reading papers vs. thinking
You don’t get ideas just by reading top papers.
If problem is obvious from paper, the authors would have solved it first!
Read papers to get an idea of the field, tools, techniques
But the core idea must come from you

6. How to get good ideas?
Start small, experiment, get hands dirty, explore
Your own experience of problems
Generate lots of ideas, brainstorm
Track latest developments, tech news
Take risks, be ok to fail
If no chance of failure, then idea is too trivial

7. Breadth vs. depth Breadth is often neglected at expense of depth
Breadth needed to find a good research problem
Make connections across areas, think out of the box
Breadth also ensures you can evolve and adapt with changes in the field
Once you zero in on a good problem, focus on depth
Depth in a few topics, but always retain breadth
Jack of all trades, master of a few!

8. Practical constraints
Align your work with your skills and interests
Also with your advisor’s skills and career goals
Did you do well in courses in that area?
Pick problems that can be satisfactorily solved and evaluated within constraints of academia
Are datasets easily available?
Is equipment to test your idea available?
Is funding available?