Some thoughts for prospective Ph.D. students

If you are a Ph.D. student and would like to work with me please read the following before contact.

Your Motivation.

It is important for you and me to understand what your main driving factor behind going for a Ph.D. is. Is it money ? Is it fame ? Is it the inability to do anything else ? Afraid of real life ? The fun of research ? Coming to America ? Any of these could be a valid answer. The true answer is usually somewhere in between, a combination of factors. It is important to match this drive with your final goals. While it is possible to do, a Ph.D. can be of very little benefit and success if completed for the wrong reasons. All things considered, a MS degree is often going to offer more money (considering the years spent) and 3 extra years in your life doing nice things in the real world instead of toiling away at research in a lab, for peanuts. Think about it !

Initial Interaction.

I would prefer to have an initial independent study course interaction before entering into deeper things. It is helpful to see if we can work well together. This also means that I will generally not provide RA support during this independent study. I'll wait for evidence that we can obtain research results at a reasonable level. As an alternative to a course taken with me, you may be able to use credentials from other courses you took, e.g., if you took other security, OS or networking courses and did brilliant.

When taking a student the intentions are to graduate him/her successfully. It is likely that this is happening so don't worry about anything else but having fun doing research and having a life in the meantime too. I expect you to do research and publish. At the conclusion of this initial interaction it would be great to produce a tech report together (e.g. in a 12-15 page Oakland/SOSP/SIGMOD publication format) of a publish-able research idea, reasonably close to submission to a 20-25% (we are lowering that for the next papers if we decide to start working together) acceptance rate conference. I intend to invest time in you to get you through basics of writing and publishing. I want you to succeed. I should be offering you an RA as soon as I can once we decided to work together (or work with you to get appropriate RA funding asap). Please do let me know as soon as possible if you don't have a laptop and/or need additional resources. I may be able to help.

Research Area.

Research-wise I do not want to duplicate research my colleagues are doing. If you are in the process of looking for an advisor I strongly encourage you to go and talk to all the people that you find you might be able to work with. Do not rush into decisions. It is not a trivial task to find a good fit and it is almost impossible to find the perfect one. Sometimes being co-advised by two people just does it. I had two advisors. While in my case things worked out great, I would not recommend it as a general technique. A fresh incoming student usually has little (if any) clue what exactly her/his thesis is. This is natural and comes with time once you start to do background reading and look around. This is why it is important to start your research by reading, reading, reading. And listening. It is good for you to go to all research talks that you can attend, even if only to see the speaker. It often helps to get new insights into your own area. Never discard a talk just because it is in a different area.


As a junior student, part of your background reading cycle is reading at least 2 papers per week, preferably well cited research in your area. It would be helpful for you to give one informal research presentation per month of something you read. You shouldn't spend more than 90 minutes and 10 slides to prepare it, nothing fancy. The presentation is to take 30 minutes sharp including 5 minutes for questions. As a senior (all quals done, at least 4 semesters old in the department) student, you are to read a minimum of one well cited or new conference (sub 15-20% acceptance rate) research paper per week. I also intend to out-source some of my less demanding peer reviewing work. This is helpful for both your resume build-up and your reading. I intend to spend some time to teach you how to review papers.


I prefer us to use (i) Linux x86 for experimental research and latex for research writing (I am willing to spend time to teach you unix and latex basics) (ii) MS powerpoint/windows for presentations. (iii) xfig and/or something that produces vectorial EPS files for figure drawings (no MSPaint). It would be nice and helpful for you to have a pencil and a paper notebook where you write down research ideas. There is a world of wood/paper out there, much healthier for your eyes and much more portable. Try to type in something research-related every day. Even if just 5 lines for a research paper or 20 lines of code for an experiment. Have always 2 things in the pipeline. One that is of high priority and one for the future that gets your attention maybe only once every 3-4 days for a few hours. One thing that I found helpful is to always bind a piece of research to a major, maximum 15-20% acceptance rate conference deadline within the next 3-5 months (this might not work for bootstrapping longer-term systems work, but works fine for progress steps, once things got started). The deadline itself and the associated pressure to make it often provides a great incentive against procrastination during those lazy weekend afternoons. More on this later.


Please create a web page in the first month after we decide to work together. It should contain at least a (more or less professional looking) picture of you, a list of the papers and links to projects you are working on, invited talks you gave, (and other research results) conferences you are going to, a link to our group and some personal information. Please update your web page at least twice a semester. It can be extremely important for you in the long run. It is definitely helpful for you to know what all of us in the group are doing (roughly of course). I support you to go and present most of the research results where you are a first author. This is why english (see below) is very important. I am going to try to help you go to all selective conferences (15-20% or lower acceptance rate) where you have a paper. I expect you to actively look for venues to give talks about your research. Please arrange to have a talk at least once a semester, even if only in a department seminar. Internal group presentations don't count. It would also be helpful for you to know at least 3-4 other students in the department (outside of our group) well enough to make reasonably well informed statements about their research. Unless it is directly research related I do not encourage you to join any committees and/or additional organizatorics. They take up allot of time and give you the illusion of achievement. In our case achievement is measured in research results. Once you get a tenure-track position you will have plenty of time to spend in committee work :)

Socializing: Get a Life !

You are invited to a bi-weekly group lunch. You are welcome to suggest any place we should go to. I am going to try to pay once a month for lunch (but you are expected to join them all :) You should bring your significant other or a friend. Lunch happens outside the department and all dissertation research discussions are forbidden. You should join social activities in the department too. But not too many ! At least once a semester we should be going out for a social weekend day. At least one day a week you should not come to the department or sit in front of a computer. Read a book ! Watch a film ! Go out ! In the worst case go to starbucks and read a famous article (e.g. Shannon) or get a napkin and a pencil (yes a pencil) and dream of interesting research stuff, un-related to your dissertation.

Personal Issues.

I insist I know if you have personal problems that may impact your research or otherwise. I will understand if you disappear for a longer weekend without letting me know. If you leave for more than a total of 3 working days please make me aware of it in advance. You don't have to be here physically to do research and I am aware that research can happen also with a laptop on a beach, but the probability that it does tends asymptotically to zero. So if you plan on visiting your Californian boy/girlfriend starting Friday morning and coming back Tuesday night that is fine (assuming that you make all the regular scheduled meetings or let me know in advance). If you plan on going to Hawaii for the week please let me know. Remember I don't need to see you every day but I want to make sure that I put just enough pressure on you to result in sustained progress, an antibiotic for the procrastination bacteria. Sustained progress can be sometimes quantified by something like "at least 2 pages per week written on a research paper" but this is by no means a clear metric.


It is definitely important for you to speak english reasonably well if not fluent. I am not sure yet how to evaluate this objectively, but you have to be able to write a composition of my choice with fewer than X spelling errors, Y syntax errors and Z grammar errors. If necessary, I would like to help you to get to a level that is appropriate for a Ph.D. by the time you graduate (e.g. by recommending english classes and books to read). I would be reluctant to graduate you before this happens. It is important for you.

Schedule and Timeline.

Besides the timed items appearing elsewhere in this document I prefer we have a personal meeting once a week or more for fresh students. Once every other week for senior students. Group meeting once every two months. All of these are scheduled in the first week of each semester. You graduate in a minimum of 4 years if going for industry, 4.75 years if going for academia. Note the word "minimum". It is very bad for you to rush it. I almost did and it was not pretty. I remember in my graduate student years the urban legend was that professors keep students around longer to "exploit" them. As time passed by I understood that this is often the opposite. Although I was told I am ready, it turned out to be in my own best interest to not graduate after 4 years only but rather wait one more year and get significant more research done. However, with very few exceptions, I believe that you should graduate in no more than 5.5-6 years.


Please read the following guidelines on ethics: Professional Ethics in CS, The IEEE Code of Ethics, ACM Code of Ethics and Professional Conduct.

Your Expectations.

I am to be available within our meeting schedule and mostly also outside it. I am to help you graduate successfully in a timely manner. I am to help you to the best of my abilities to get a job if you do your part, but this is in no way a guarantee that you will get one. The market and a million other factors influence that.

Your Input.

Please feel free to always suggest any improvements to this document, or to our interaction model. It is essential you do so and we diffuse any potential things that don't look right in your vs. my model of the world. Do not forget that I want you to succeed 110%. We have to synchronize our models ideally to work best and communicate.


This document can be subject to changes. Please check from time to time.

Last updated: December 23, 2009