UCSB Computer Science
The following are personal opinions and observations (based on a modified version of a presentation by H. T. Kung at CMU in 1987). These are not department or university policy.
The PhD process is sort of an apprenticeship. You're not there just to do a thesis; you're there to become a mature and knowledgeable researcher in your field. Sometimes this means working on non-thesis-related projects for a time, helping out with lab/group duties, giving lectures, giving demos to various people, helping the advisor with reviewing or proposal writing, etc. etc. All of these "distractions" can be quite useful to you later on.
The PhD process is long and arduous. Many (most?) people consider quitting at some point. It helps to have clear goals at the beginning, to remind yourself while in the midst: Why do I want to do this? Am I committed to persevering through it?
What's a PhD thesis all about?
A PhD thesis represents a substantial body of work. It should be marked by high quality and substantive results. It should push the frontier of knowledge. It should mark you as an expert in your area.
Thesis research is partly intended to ensure that the student can later take on independent, long-term research commitments.
Faculty are judged by the theses of their PhD students
High quality PhD theses is one of the most important factors contributing to the success of leading universities
PhD thesis research is a big challenge. It is difficult. There is no simple formula for success. There is no clear standard for a PhD thesis
Typical stages of graduate student life (in chronological order):
Knowing everything (A little knowledge is a dangerous thing)
Knowing nothing (I am not worthy)
Gaining confidence (Hey, I can do this!)
Knowledgeable and confident (Based on real experience)
Here are some things a PhD thesis might do:
Open up a new area
Provide a unifying framework
Resolve a long-standing question
Thoroughly explore an area
Contradict existing knowledge
Experimentally validate a theory
Produce an ambitious system
Provide empirical data
Derive superior algorithms
Develop new methodology
Develop a new tool
Produce a negative result
Your advisor does not know everything. Your advisor may not always be right. Your advisor's knowledge in some areas may be obsolete. Remember, the only one who is ultimately responsible for your progress and success is you.
On the other hand, most likely your advisor does know a thing or two, so don't ignore or avoid him/her.
Advisors are real people, and they may do crazy things like change jobs, change research interests, or be denied tenure. Don't despair, there are usually ways to work these things out.
If at all possible, keep a good and open relationship with your advisor. Your advisor usually has lots of influence even after you graduate. He or she should be genuinely interested in your well-being, and may even be a mentor for your entire career. Mostly likely, your advisor will be writing letters of recommendation for you - if they are not strong, that can hurt. Despite the pressure to always make things look good to your advisor, the best course of action is to be open and honest at all times.
When you have the chance, schmooze! Get to know people in your field: the faculty in your department, other grad students at your university (in your department and others), other grad students from other universities, faculty and researchers from all over (especially those whose work you know and respect). Don't just wait to be introduced (although your advisor can certainly help here) - introduce yourself. Talk with visitors to your department, aggressively meet people at conferences. You never know who will have a job for you, or ask you to be on a good program committee, or play some other helpful role sometime in the future.
Consider your fellow graduate students as a very valuable resource - ask them questions, see how they do things, tell them your ideas, brainstorm with them, learn from them! Don't just hang out on your own, or you will miss out on a lot, both personally and professionally.
Think. Don't just do. Set aside time to explore ideas. Thinking is more important than reading.
On the other hand, don't procrastinate. Set aside time to code, experiment, build, read, or whatever. Don't think you have to understand everything before you can begin to make progress.
Early in your PhD "career" you will be taking courses, jumping through various hoops (like a qualifying exam), maybe helping with other, unrelated projects, and perhaps working summers in industry to get some valuable experience (and a little cash). Later in your PhD career, you should be focused on your research. Learn to say no to distractions, especially after the first year or two.
Stages of PhD thesis research
Selection of the area
Selection of the advisor
Becoming a researcher in the area
Building up general knowledge and experience
Learning the important issues and questions in the field
Learning the cutting edge work in some areas
Some useful things to do:
Read recent proceedings of the best conferences, and ask more senior people what were the best papers. Try to figure out what makes a great paper (and thus what makes great research).
Keep a notebook that contains your research notes. Put all of your empirical data and initial ideas in the notebook. Make notes on a paper as you read it and think about the assumptions of the author and the importance of the results.
Follow references from one paper to another until you know an area extremely well. Don't count on your advisor to hand you all of the relevant papers out of his file drawer. He/she doesn't have them all!
Build a mental model of what has been done in your area. Look for holes - interesting areas that have not received much attention.
Most crucial stage, since everything flow from here. Later problems can often be traced back to a weak thesis proposal.
This is where you need your advisor the most
Main challenge: come up with an approach and/or an experiment.
Don't just go with your advisor's opinions or recommendations (unless you truly believe them). This is the best time to argue with your advisor!
Research plan - Overview of the expected course of research. Must be flexible, but not vague.
Need to elaborate on the focus, the approach, experiments or systems to build, potential impact
Forming a committee - Choose people who can help with needed expertise if possible, especially if you have an interdisciplinary topic.
Thesis proposal questions:
What is your approach and what is new?
What is your secret weapon?
How do you measure your own progress?
What are the success or completion criteria?
How will the expected results change the-state-of-the-art?
Be honest - don't exaggerate your claims, be open with the weaknesses (better for you to raise them than for someone else)
Pick a problem/project of manageable size. It is much better to do an excellent job on a moderately sized project than a moderate job on a large project.
Prepare a tentative month by month schedule, with milestones, for your work. Be realistic. And flexible, but not vague.
Keep your advisor and committee informed. Don't wait until you have a breakthrough result, keep them informed regularly.
Talk about major choices with your advisor before you're completely committed to them.
Knowing when to stop
How much is enough? Your thesis does not have to solve every possible related problem. Talk with your advisor about what a reasonable stopping point might be. Re-visit the issue occasionally.
If the principles and boundaries of your thesis work were clearly defined from the beginning, this should not be a problem.
Writing is very time consuming. Really. Don't plan on "writing it up" in a hurry.
Don't spend excessive time fiddling with fonts, formatting, etc.
Do take care in producing good content - proper English with no spelling errors, typos, etc.
The document should be well organized and easy to read, not just a bunch of math or results, and not just a chronological record of what you did.
Write for an audience educated in your general field, but not just for an expert in the particular subfield.
Motivate the reader - make sure the reader understands the problem, why it's interesting and relevant. What's the Big Picture?
Remember, few (if any) readers will have your background. Don't assume they know everything you do.
Write the introduction last (or at least re-write it last).
Comments from the committee
Committee members are very busy people. Don't expect them to go through several versions of your dissertation.
You want to give them an optimal version for them to comment on some time before the final version - not too preliminary, not too finished.
This should mostly be a formality, at least from the advisor's point of view. That is, he or she shouldn't let you get to this point if you're not really ready. There should be no surprises for the committee.
However, it's still important:
The defense gives you a chance to get feedback for final improvements to the thesis.
Many people (the committee and the audience) may form their opinions of you and your work from this one event
Presentation material can be used for future presentations (job talks, etc.)
You should know firmly what the main ideas are, and present them clearly.
Set the defense date well in advance - it can be difficult to get the committee together.
Usually there is some work to be done (often minor) requested by the committee
Publications: conferences, journals, book
Are you the sole author of these publications, or is your advisor a co-author (even though you wrote it)?
No single answer, but typically the advisor has been very involved and deserves credit of authorship. Ideally, he or she will also help with writing (and re-writing, and proofreading) the subsequent papers.
Follow-on work - Are there more ideas or paths to take on this problem? You're now the expert, so why not keep working on it? Starting over on something completely different is not so easy, especially if you're going to be looking for research funding.
Take a vacation - you earned it!