During the semester you will be writing a term paper that will afford you the opportunity to study an operating systems topic in detail and to present the results of your study in a written form. The paper is intended to be an academic paper, similar to the ones that we discuss in class, although the paper is not required to be of publication quality.

There are two major goals of the paper, and they are the basis on which the paper will be graded. I want to see that you have:

This is a class, and I don't expect each student's new insight to be a field-changing idea. It can be as simple as a new way of organizing the major ideas in the field so that they make more sense, or a gap in the assumptions of the practitioners of the area that is promising for new research.

The goal is to encourage each student to have the experience of researching an area and understanding it to the point where their understanding can yield new insight. As you pick your topic and compose your paper, keep these goals in mind.

The writing process consists of a proposal, which must be accepted before 5 Apr 2009, and writing the paper itself. The proposal outlines the topic and scope of the paper, and we will use a formal proposal system this year. Once the proposal is accepted, a full term paper must be produced by the end of the semester. That ordering is not intended to indicate that you cannot begin working on your paper before your proposal is accepted, but both hurdles must be cleared according to their own deadlines.

For many of you, this will be your first academic paper. There are three aspects to this process: research, understanding, and writing. The first step, research, consists of exploring your topic by reading the published literature in the area. Your goal is to become expert on a topic, so you'll need to explore this area extensively. You will need to read and understand papers in the area, and develop an understanding of trends in the area. As you read more, the process of assimilating new papers will get easier, but there is a learning curve to master. You will be looking for the most influential ideas and how that has affected research in the area. Do not hesitate to explore a field's past or recent publications. There are papers on worms and viruses from the 1980's and on hypertext from the 1950's. Both of those areas are active areas of publication today.

Finding those papers is the first step, and can be a difficult one. Web search engines are a fine technology for getting a foothold in the literature and finding those first few strong papers one needs to cover a topic. As you find sources on the web, be sure to view the sources critically. Papers that have been peer-reviewed and published in journals with solid reputations are best. Some strong Operating Systems journals include the ACM Symposium on Operating Systems Principles (SOSP), the USENIX Symposium on Operating Systems Principles and Design (OSDI), and any number of USENIX-sponsored conferences. Beware of unpublished white papers and marketing hype that has minimal technical content or review. Also note that ACM SIGOPS's Operating Systems Review (OSR) is not peer-reviewed. If you have questions about a conference or journal's reputation or a source's suitability, feel free to contact me.

Another method of finding literature on a given topic is to scan the tables of contents of the proceedings and publications discussed above, either electronically or by going physically to the library. This can be extremely valuable if you have a broad area of interest that you would like to narrow down or if you want to know what the major subfields of a given area are. For example, someone interested in security could find out a lot about the directions of security research by looking at the contents of the last several proceedings of the USENIX Security Symposium.

Once you have found a paper or two, the references in those papers may also be helpful in finding more papers. Hopefully you will begin to see patterns in the references, noting which papers are seminal and what sub areas are important.

Once you have gathered and read enough to begin understanding an area, you need to begin working on structuring the knowledge you have. Some good questions to ask at this time are:

Any one of these questions can lead to an insight that may help you better understand the area and even suggest new directions in it.

Once you decide what principles will organize your paper, even fairly prosaic ones, you need to begin designing the paper. You have reached some conclusions or gained some understanding, and now need to transfer that to me, your reader. You need to present what you have learned logically so that I can share your ideas. I may not agree with them, but I want to understand what they are and where they came from. I may also agree with them completely.

Start by asking what conclusions you have reached. Write a couple sentences saying where your thoughts have ended up. Now work backward to help your reader reach those conclusions. Arrange what you have learned to lead one to those conclusions, including evidence that supports your conclusions and any that detracts from it.

You will probably want to start by explaining (some of) the systems you have studied, their goals and function. You will also need to discuss how they are related and what ideas are embodied by them. It may be appropriate to talk about the environments each was designed for and how well they met their goals in that environment. Your goal is to let the reader understand the ideas and show how your conclusions naturally follow.

The design of the paper can take many forms; an outline is a common useful format, but different people's minds work differently. You will not be graded on any outlines or other organizing documents you produce, but I encourage you to produce them. Having a plan and sticking to it can significantly improve your paper. Depending on how comfortable you are with writing, you may want to outline the paper into sections or even right down to the paragraph level.

Once you know what you want to say and how to say it, writing the actual text is time-consuming, but straightforward. Write clearly, simply, and avoid the temptation to create suspense. By the end of the introduction, a reader should know the basic conclusions that the paper will defend and the basic outline of the argument. The rest of the paper should supply background, lay out the specific details of interest and make the relationships between those details clear, leading the reader to the conclusions.

Once you have completed a draft, let someone else read it and give you feedback. If possible, find a native speaker or strong writer to do so. You are allowed to take that person's input without fear of violating the academic integrity requirements. Use good sense, though - this person may not edit your paper or provide you with notes from their own research, for example. If you have questions about if an particular action is acceptable, contact me and confirm that it is acceptable.

If it seems like this is a time consuming process, I have given you the correct impression. Set aside time to do the work; do not expect to turn in a paper written overnight and expect a good grade. Research should be done steadily throughout the semester and could take 2 weeks or more of reasonable attention - say 2 hours a day on every other day. The outlining step can take another couple days of 2 hour effort. Finally writing the paper can take several days of concentrated effort, more if you are not a proficient technical writer. Do not be surprised to spend a week. Those numbers are just guidelines, you may be slower or faster.

You may discuss your outline or topic coverage with me, and in general terms with other students. I will not read your full paper other than to grade it, however. You can show me paragraphs or excerpts if you have questions. If you have questions about proper citation of work or plagiarism, please bring those questions to me. I will read larger excerpts for the purposes of avoiding plagiarism. Sooner is much better than later for such questions.

It should go without saying that the term paper is subject to the USC code of Student Conduct. In particular, should a paper be plagiarized, the student who “wrote” it will be subject to the harshest penalties. The least that will happen to you if you are caught violating the code is that you will fail the course.

There are many resources that can guide you in avoiding plagiarism. They include the Office of Judicial Affairs and Community Standards' publications, especially Trojan Integrity: A Guide for Avoiding Plagiarism; speaking to the people in the Office of Student Judicial Affairs and Community Standards; and speaking to me.

Even if you are certain you understand what plagiarism is, I encourage you review the documents above. Ignorance will not be accepted as an excuse for plagiarism; it is your responsibility to know what plagiarism is and not to commit it.

Bear in mind that even general reference works such as Wikipedia and encyclopedias must be properly cited if their words are used verbatim. If this is a surprise, that is an indication that you need to read and understand the plagiarism information referenced above.

Also bear in mind that the proposal, though informal, is subject to the same academic integrity standards as the final paper. A plagiarized proposal is grounds to fail the class.

Despite the repeated warnings to the contrary, every semester students plagiarize papers or proposals. Some get caught. Without exception those caught fail the class. Do not expect to talk your way out of an F in csci555 if you are caught plagiarizing a paper.

There are two acceptable formats for the paper:

Text beyond the prescribed maximums will be ignored.

Papers must be submitted electronically to <csci555@usc.edu>. Postscript, PDF, and ASCII are acceptable formats. If you submit Postscript or PDF, you need to get a confirmation from me that the paper can be printed. Your submission is not complete until you get such a confirmation. Papers received on time, but not confirmed until later will incur no penalty.

It is wise to send a short submission from your document preparation system to me well in advance of the paper due date to confirm that your software can produce printable output and that you are capable of doing so. I'm happy to provide this feedback. Debugging remote printer problems under a deadline will prematurely age you.

All deadlines are at 11:59:59 PM Pacific time on the date given - the last second of the day. It is unwise to push those limits if it can be avoided. One example of a potential problem is that I will not necessarily be reading my e-mail until midnight on a due date to let you know if there is a problem.

Missing any of the deadlines above means your paper is subject to the late penalties outlined below. In particular late submissions or acceptances of proposals will reduce your final paper grade.

A late paper is subject to the following penalties:

I will not accept papers more than 5 days late.