On this page:
1 What is a Ph  D
2 1,000,000 and a few reasons why you don’t want a Ph  D
3 Why would you want a Ph  D
4 What is the training like
5 How do I find this master
8.1.0.7

Why/How do I get a PhD

Matthias Felleisen

1 April 2021, 5 January 2019, 14 November 2012

Undergraduates often ask me what they should do to get into a good PhD program. So here are my thoughts, some of my own, some based on what others told me. All of it is based on my own personal experience of how I stumbled into a PhD program, and how I have trained PhD students for the past 25 years.

Changed in version 1.3: Thu Apr 1 09:13:59 EDT 2021 adds a section on why you would not want a PhD, plus some touch-ups

Changed in version 1.2: Sat Jan 5 12:25:44 EST 2019 adds a section on why you would want a PhD

Changed in version 1.1: Wed Aug 3 21:26:51 EDT 2016, detailed feedback from
    Celeste Hollenbeck on my writing. Thank you.

Changed in version 1.0: 2012: initial release

1 What is a PhD

Simply put, a PhD is just a piece of paper. Some need it so they can sign documents as "Dr. SoAndSo," and if this is your goal, this document is not for you.

Others have figured out that undergraduate courses are fun but working on questions independently and figuring out new insights is even more fun. That insight is the first step toward a research-oriented PhD program. Because such programs are the place where you learn to pursue such independent inquiries and where you get the credentials to do so for the rest of your life—if you so wish.

In short, a PhD is a ticket to a life of research and exploration, a life of learning for learning’s sake, and a life of sharing the joy of discovery with like-minded people.

2 1,000,000 and a few reasons why you don’t want a PhD

Some recent graduates start with salaries in the $150,000 range, give or take $50,000. These numbers are based on reports from undergraduates that I have worked with over the past few years. Graduates who choose to work for one of the biggies can earn over $200,000 in a couple of years.

On the average, obtaining a PhD in computer science takes five to six years. The average starting salary for a freshly minted PhD in academia may equal that of a software engineer.

Now do the math. Getting a PhD is not about earning power. If that’s what you’re thinking, stop reading here. As for the other “few” reason, I will add those in a few years from now.

3 Why would you want a PhD

Obtaining a PhD enables you to apply for positions usually out of reach for those with just a bachelor or masters degree:
  1. a tenure-track (research) university level professor

  2. a tenure-track (teaching) college professor

  3. a (usually) non-tenure track university lecturer, recently upgraded to “teaching XYZ professor” and similarly titles at many universities

  4. a high school teacher at a high-priced private school where students are prepared for “elite” college entrance

  5. an industry researcher

    Some companies still have research labs where kind-of-academic research happens: IBM, Microsoft. Galois in Portland is a company dedicated to (contract) research.

  6. an advanced researcher-developer

    Google X would be an example, contrary to the company’s “moonshot” propaganda, which may make it sound like an academic research lab.

  7. a super-hacker (someone who solves hard coding problems)

  8. an ordinary developer whose PhD counts for “experience” (but only if you built software systems during your PhD)

  9. a team leader/manager

  10. a start-up founder because your PhD work has financial potential

    (You have read too many of Paul Graham’s essay.)

  11. a diplomat (but don’t work with me then :).

This list describes only a range of the possibilities; the last bullet indicates that you can also leave behind your computing background and apply your newly acquired problem-solving skills to other domains.

Some of these positions may pay off financially in the end; but, see above.

In other words, there really is no need to get a PhD other than that:
  • you are willing to give up a substantial amount of income

  • you love to learn Have you ever fallen in love? is how I always open a conversation on this topic.

  • you want to be with others who share this attitude

  • you want to go for an “academic” research or teaching position (bullets 1, 2, 3, and 5)

Is it worth becoming an academic researcher or teacher? Ask me

4 What is the training like

A PhD student is an apprentice to a master, also called a PhD advisor.

In this day and age, a PhD student is often also a member of his master’s team or lab or "shop."

The PhD advisor has a number of tasks. In science and engineering, an advisor’s most important task is to raise funds for his PhD students. He is responsible for getting enough money so that a student has bread, water, and shelter. On some occasion, funding may involve teaching, but every PhD student should have a few years of research time. Teaching is a privilege, and learning to teach is an essential part of learning to do good research. Ask me.

In return, the PhD student and the advisor usually collaborate on a jointly developed research topic. Indeed, the advisor’s second task is to create a "playground" where the PhD student can find a dissertation topic. Some advisors have a list of ready-made topics; others have a playground and work with the students to find a mutually agreeable topic. I also know of advisors who provide little or no input. These professors are advisors in name only.

Once the advisor and his apprentice have found a theme, they embark on the research process itself. It is the student’s task to formulate questions and to come up with solutions, though the advisor must pruneTo this end, an advisor must know the literature well enough to know what people tried before, why these attempts failed, and whether a different one may fail for the same reason. Otherwise history repeats itself. obviously dead-end explorations and should on occasion inject ideas. In the first few years, the advisor will contribute numerous ideas, possibly more than the student. Over the years, however, the advisor must pull back and must allow the student to take possession of the topic.

Pragmatically, the process also involves tedious labor: learning to read papers, learning to articulate the questions, figuring out solutions, convincing others that these are solutions worth their attention—both in writing and in conversations or presentations. From the perspective of cost accounting, this part accounts for about half the process or more, simply because articulation skills are critical for your future as a researcher. No matter what you end up doing—research at a research university, research in an industrial lab, teaching, "super-hacking"—as a future PhD, you will have to communicate ideas. Communication includes the ability to speak informally about his work, to give formal presentations, to lecture about it in a classroom, to write papers (memos), to "teach" colleagues without teaching them openly, and so on.

Make sure your potential advisor is willing to spend a lot of energy on this part of the process. This is where teaching comes in, and a good advsior works with you on teaching well.

5 How do I find this master

Given the preceding, it should be clear by now that you should not try to get into a big-named school.

The goal is to find an advisor that fits your interests and personality.

Such advisors may work at schools with big names. They may not. Today you can find good advisors at all levels. If you can find a school with a range of advisors, great. Here is how you go about it.

The first step is to figure out which kind of questions interest you personally and which ones don’t. Don’t confuse the interest that a great teacher generates with your interest. On occasion, these two coincide; often the former misleads you to think it is or at least should be your interest.

If you are not a senior yet, find someone in your department in your area of interest. See whether you can work with him for a summer. Researchers often have money for introducing undergraduates into their research areas. Even if such monies aren’t available,... or if a university sets a minimum wage that prices undergraduates out of the market ... you might be able to work as a volunteer on some small projects. If such a project can lead to a paper, grab it and contribute as much as you can. A paper will help you grab the attention of good advisors. And participating in a paper-producing process will tell you whether you really find this area interesting.

The second step is to find the top conferences in which people publish results in this area. Sit down with the conference proceedings and read the abstracts and introductions of as many articles as you can. Start with the most recent ones and work back five to ten years. Keep a list of the “senior” authors who write the papers that interest you.

Once you have a list of such “senior” people, research them. Where are they now? Do they still work on the kinds of questions that interest you? What are they working on now? Do they have PhD students? Where did their PhD students end up?

You may even wish to contact them. If you do, don’t bother with generic emails. Write a message that lets them know that you have read their papers, that you know their current research projects, and that you know which tools they use to conduct research.

The third step is to apply to the five places (or so) that have advisors in your area of interest. Visit the advisors who accepted you and find out what these people are like in person, how their students work with them and each other, what the atmosphere is like in their labs.

Now pick the one that is a great match in terms of interests and an equally good match on the personal level.

Good luck.