Notes from a lecture given by Professor H. A. Simon to first year Computer Science graduate students at CMU in the Fall of 1986.

How to do research:

1. A problem is a question, not a topic.

What would be an answer?

What would you like to know when the research is over?

How would you know when you succeeded?

2. A good problem is one which:

Someone cares about the answer

The present state-of-the-art makes research feasible

3. Where do problems come from?

Real-world issues

Scientific literature

• Be familiar with it
• Remember that everyone else reads it too
• Avoid faddish research

4. Solving a problem

All reports are progress reports (so publish now. NOT when all the research is completed)

Every good problem leads to more problems . . . never ending.

5. Planning

Vagueness is inherent in research: Two items should be clear

(a) Clear criteria for what an answer will look like

(b) Have a plan for what to do tomorrow (that is, exactly what you plan to accomplish in the next 24 hours)