The master's takes you where others have been; the doctorate, where no one has gone before. (New York Times, June 6, 2004)

In this short article, the pre-thesis requirements of Computer Science Ph.D. programs at various departments are compared with those at National Taiwan University. The following world-famous institutions are selected: U.C., Berkeley, MIT, CMU, Princeton, and Stanford. By pre-thesis requirements is meant the requirements that must be met before a thesis work can commence or that are mostly orthogonal to the thesis requirement such as proposal, dissertation, and its defense thereof. At the end, a comprehensive table summarizing the findings is presented for a quick reference. The conclusion from that table seems clear: Our department places the fewest demands on the students.

A minimum of 18 units is required. A student must pass comprehensive exams in 4 areas out of 6 by the 5th semester. The 6 areas are Algorithms, Architecture, Artificial Intelligence, Compiler and Programming Languages, Operating Systems, and Theory. Among them Theory is mandatory.

A minimum of 32 units is required. (One unit means one hour of lecture per week.) Among them, there must be 16+ units of graduate courses for EECS Major with a GPA of 3.5 better and 8+ units of graduate courses for EECS Minor with a GPA of 3.0 better. The CS breadth requirement consists of 3 courses, each chosen from the list of courses in Systems, Theory, and Theory+Systems. The grades for these courses must average at least 3.5, and only one course with a minimum grade of B+ will be allowed.

A Ph.D. student with a Master’s degree in Computer Science must complete the requirements by the end of the 3rd semester. One without a Master’s degree in Computer science must complete the requirements by the end of the 4th semester.

If a Ph.D. student already has a Master’s degree in CS, he needs to sign up for the oral prelim exam by the 2nd semester of his Ph.D. program and complete all preliminary requirements (oral and breadth courses) by the end of your 3rd semester. He needs to score at least a 6.0 out of 10.0 to pass. The areas are Artificial Intelligence, Architecture, Computer-Aided Design, Database Management Systems, Graphics, Human-Computer Interaction, Networking, Numerical Methods, Operating Systems, Parallel Processing, Performance Evaluation, Programming Languages, and Theory.

There is no explicit breadth or depth requirement. A Ph.D. student must complete 96 university units worth of elective graduate courses for breadth across many areas in and beyond Computer Science. Five of them must be star courses, one per five areas in Algorithms and Complexity, Artificial Intelligence, Computer Systems, Programming Languages, and Software Systems. One way to obtain elective unit credit is through a faculty-directed project. Another way to get elective unit credit is by being a teaching assistant for one of the Ph.D. star courses.

A Ph.D. should complete 66 units of coursework.

The Technical Qualifying Examination (TQE) requires students to demonstrate competence in four different technical topics. Competence in each topic can be demonstrated by taking a graduate subject designated for the topic (see below), and earning at least an A-. If a student gets two or more grades less than A-, an oral examination will be required on those topics. A grade less than B- in a single topic also requires an oral examination. Students should complete all components of the TQE by the end of their 2nd year in residence.

There are 6 topics, two in each of Systems, Theory, and Artificial Intelligence. Students must choose one topic from each group (Systems, Theory, or Artificial Intelligence). The fourth topic may be in any group.

A Ph.D. student takes 6 courses to satisfy the requirements. A grade of A- or higher will normally be expected in the required courses. They fall into the areas of Computer Systems, Software Systems, and Theory. The requirements must be completed by the end of the 2nd year. (Taking a written examination given for all first year students each January and passing it is an alternative.)

First, each student must register for 135 course units for graduation. CS Ph.D. students take 9 units (courses/research) a quarter. Second, students must take courses from at least four different faculty members. There are no courses specifically required by the Computer Science Department.

Then there are comprehensive examinations in core areas for breadth requirement. Each area has associated with it a set of readings defining the required material for the area and a course(s) alternative.

In order to satisfy a given part, a student may pass the corresponding portion of the written comprehensive examination (offered once each year) or satisfy the course(s) requirement for the area with a grade of A- or better. This year there will be 12 comprehensive areas offered with corresponding exams and/or course work. However, each student only needs to pass 8 of the areas.

A student must pass five whole areas by the end of the 2nd year for reasonable progress and must pass 8 of the areas by the end of the 3rd year in the program. A good strategy, in fact, is to take all of the parts of the examination in the first quarter and then take courses to satisfy any parts that were missed. A student must pass 6 of their chosen 8 area comprehensive exams to be eligible to file for candidacy.

The format of the written comprehensive exam is a series of one hour tests scheduled over six days (two tests per day) so that a student can take any or all of the tests in one offering. The questions in each area are often formulated by the professors in charge of the courses for that area (or by individuals designated by those professors).

It is a University examination. The purpose is to demonstrate that “the student is clearly an expert in those areas of the discipline that have been specified for the examination, and that he or she can, in all likelihood, design and produce an acceptable dissertation.” Very few students fail. The exam is meant to demonstrate readiness to do research; it is not intended as a defense of an all-but-completed dissertation. All reasonable effort should be made to take the exam in the third year.

The Research Qualifying Examination (RQE) monitors students' research progress as well as skills in presentation, both written and oral. Students should aim to complete the RQE by the end of their 2nd year in residence.

The RQE is normally taken on or near completion of a Master's research project or comparable research experience---at the end of the third graduate term, and in any case not later than the end of the 4th graduate term. (Master of Science is required of students pursuing a doctoral degree at MIT.) A conference-style (less than 20 double-spaced pages) paper based on original research by the student (usually the Master’s thesis). The RQE Committee conducts an oral examination in which the student is asked to present his/her research and to defend it in discussion.

The examination is meant to show that the student is actively engaged in a program of research and has acquired all the appropriate technical background. It consists of a research seminar (open to the public and advertised in advance), followed by an oral examination. The oral tests the student's knowledge in a number of topics relevant to the student's research area. It must be passed by fall of the 3rd year. The area exam may be taken twice. Failing the exam a second time implies automatic termination of degree candidacy, but conditional passes are possible. The precise outcome of a conditional pass is determined by the student's performance in the exam as well as any other information the faculty might have to help identify the student's strengths and weaknesses.

A student may take the qualifying examinations only twice. Often a “conditional pass” is awarded, and when the designated conditions have been met (such as teaching a certain class, taking a course, or reading additional material in a specific area), then the student would be credited with the Pass. The format of the qualifying examinations varies from year to year and area to area. Recently, there have been in-class written exams, “take-home” written exams, oral exams and writing assignments. Within the first 3 years, a student must pass a qualifying examination in the area of his/her intended dissertation.

During a student's first two years, he should be doing directed research at least halftime. Once all coursework is completed and before doing thesis research, he should be doing directed research full time (except when teaching).

At each semi-annual faculty meeting, the faculty review the student's previous semester's research progress and the student's next semester's research plans to ensure that the student is making satisfactory progress. The evaluation of a student's progress in directed research often depends on the student having produced some tangible result. Advisors are individually responsible for adequately supervising this portion of the Ph.D. program.

Each student should have the abilities to communicate technical ideas clearly in writing and to communicate technical ideas clearly orally.

To satisfy the written communication skill requirement each student must write a scholarly document, as either its sole author or its primary author, that is at least the quality of a Carnegie Mellon technical report. The student obtains written final approval of the document from at least two faculty members and one graduate student. We expect students to be able to satisfy this requirement within their first three years and prior to their thesis proposal.

To satisfy the oral communication skill requirement each student should give a public talk at Carnegie Mellon. If the student does not pass the test, he may give another talk or do a “remedial.”

Each student must present a public technical lecture, reviewed by a faculty member. Most often this requirement is satisfied by a presentation in a departmental research seminar.

The pre-thesis requirements of Computer Science departments are tabulated below for a quick reference.

		NTU	UC, Berkeley	CMU	MIT	Princeton	Stanford
Courses	Load	18 units	32 units	96 units	66 units	18 units	135 units
	Breadth/depth requirements?	no	yes	yes	no?	yes	yes
	Minimum grades requirements?	no	yes	no	no?	yes	no
	Substitutions for course work?	no	no	yes (project/teaching)	no	yes (annual tests)	no
Exams	Oral preliminary exams?	no	yes	no	yes (if comprehensives failed)	no	no
	Comprehensive exams?	yes	no	no	yes	no	yes
	Substitutions for comprehensives?	N.A.	N.A.	N.A.	N.A.	N.A.	yes (course with grade A- or up)
	Qualify exams?	no	yes (oral; easy)	no	yes (oral and written)	yes (oral)	yes (various)
Misc.	Research?	no	no	yes	no	no	no
	Writing skill test?	no	no	yes	no	no	no
	Public talk skill test?	no	no	yes	no	no	yes

http://buffy.eecs.berkeley.edu/Students/Grad/Affairs/CS/phdstuds.html

http://www-2.cs.cmu.edu/csd/phd/phd.html

http://theory.lcs.mit.edu/~joanne/new-roadmap.html

http://theory.lcs.mit.edu/~joanne/rqe.html

http://www.eecs.mit.edu/grad/degrees.html

http://cs.stanford.edu/Degrees/phd-req.html