Skip to content

Phd advice

  • Maintain a productive & healthy routine and avoid burnout
  • Always know what people expect from you. Manage expectations well
  • Set your goal clearly
  • Write down everything you do — esp. lower experiments and upper understanding. Document your code
  • Organize your (virtual) workspace
  • Be honest with people and ask dumb questions. Helping your supervisor to help you is key
  • Socialize widely
  • Present your research
  • Deal with failures or divergence calmly. Take a break. Think about how to get around it.

The Red Book from Stanford - some punchlines

  1. beware of working on projects with students on the same graduation schedule as you, as you may both want first authorship on the same paper
  2. "Will this project lead to a paper? Which conference will I submit it to? Is it a chapter in my thesis?” Do not work on a project simply because it’s cool.
  3. A project will span weeks, if not months or years. It is easy to become distracted or discouraged. Set weekly goals.
  4. Ergonomic Keyboard
  5. Start meeting people early: you will only grow busier.
  6. If you do leave, try not to burn any bridges

The key to understanding collaboration and funding is: people are mostly self-interested.

So, if you are beginning a collaboration with someone, be sure that it serves a major goal, and preferably a primary goal, of your own (most often, publishing a paper on which you are first author), before investing significant effort into the collaboration.

Determine the value of something by imagining life without that something.

Sometimes it’s more important that you make a decision, than exactly what decision you make.

PhooD!!!

being self-serving in both the short and long term usually implies that you should be very nice to people overall, but without letting them walk all over you.

If people ask for too much help, make them work!

DO YOU UNDERSTAND the multifaceted, uncertain complexity?

  • You can present complex ideas to a diverse audience

Show “interesting results” to your advisor is like “pre-publication”.

  1. Turn your daily PhD work into homework assignments for yourself
  2. Work like the world’s best engineers and scientists (and not like a perfectionist)

This switching between high- and low-level thinking is critical to success during a PhD

By starting with the simplest, dumbest hypothesis possible, you can slowly learn facts about a system through a sequence of experiments and refine your initially dumb hypothesis.

Advice from mareidj

  1. The daily grind
    1. Keep a journal of your research activities and ideas, wite down 1. speculations 2. interesting problems 3. possible solutions 4. random ideas 5. reference to look up 6. notes on papers you've read 7. outlines of paper to write 8. and interesting quotes
    2. Read back through your journal periodically
  2. The divide-and-conquer strategy works on a day-to-day level. Instead of writing an entire thesis, focus on the goal of writing a chapter, section, or outline.

IDEA: a central bib + github repo for managing paper reading notes. Use Dropbox to manage PDFs. Use same filename as the bib ref name to link two.

Destructive feedback: this is bad. Constructive feedback: this is bad because and you should.

From advice for advisors:

  1. Know your students personally and help them grow in a larger aspect
    1. Sandwich the questions about principles in your detail work
  2. Find people to ask
    1. "Do you think A is interesting?"
    2. "Can you proofread my interesting proposal?"
    3. "I did A this way, what are other ways?"
    4. "Do you think this is a good results? What might threat its validity?"
  3. Find your community actively
  4. Think about how to get strong & useful recommendations
  5. Lead your interaction and ask for high-level or low-level support when appropriate
  6. Control the frequency of meeting by reflection over the effectiveness
  7. Make it casual
  8. Make it transparent
  9. Force, or at least indicate your need for written comments, suggestions, or commitments
  10. Project your high-level thought, meta-thinking, and planning in a written form and communicate with your mentor on that level as well
  11. Who you know is NOT as important as what did you do
  12. Speak out your difference in thinking and ideas 1. Chance of debate is why you come to a school rather than company at all
  13. Present a better, more comprehensive picture about how you work 1. If your advisor never visited your place
  14. Older students can't guide young ones on his/her own
  15. NO decision making meeting for unimportant things
  16. Tell, or better, present things you want to learn or learned from a course you are taking
  17. Conflicts are good. Demanding is good if both parties are clear and thoughtful.
  18. Shorten your meeting!
  19. Critical discussion of research strategy
  20. Don't give a SHIT about disciplinary boundaries, but respect each's interests and knowledge areas.
  21. Don't submit to third-rate journals. Not even second rate ones. 22.

MIT AI

Many researchers spend more than half their time reading.

the Working Papers, Memos, and Technical Reports

Reading papers:

  1. The first is to see if there's anything of interest in it at all.
  2. find the part of the paper that has the good stuff
  3. Read with a question in mind. "How can I use this?" "Does this really do what the author claims?" "What if...?"
  4. Have you read X?
  5. Get summer jobs away at other labs.

Research notebook:

  1. Record in your notebook ideas as they come up
  2. Read back over your notebook periodically

Writing down your ideas is the best way to debug them

Perfectionism can also lead to endless repolishing of a perfectly adequate paper. This is a waste of time.

Don't "sell" what you've done with big words or claims.

why it works and why it's interesting.

If you put off writing until you've done all the work, you'll lose most of the benefit.

Secret Paper Passing Network

You can only present one "idea" or "theme" in a talk. In a 20 minute or shorter talk the idea must be crystal clear and cannot have complicated associated baggage.

The rule of thumb is that any given subtask will take three times as long as you expect.

Also very important is "taste," the ability to differentiate between superficially appealing ideas and genuinely important ones