The research project is broken into four deliverables: a project proposal report (5%), a midterm project update report (10%), a final project report (30%), and a project presentation (5%). All reports must be written in LaTeX and submitted as PDFs. For teams of two, submit one document per team. Clearly identify both team members.
The proposal is one to two pages, written in LaTeX. It should clearly explain the proposed project and why it is interesting. It must include a summary of related work, the goals of the project, the final deliverables you intend to produce, a plan for making progress on a substantial chunk of the work by the end of the semester, and backup plans.
For teams of two, explain the planned split of the work.
The midterm update is two to three pages, written in LaTeX. It should explain your progress so far and present any preliminary results. It must contain a full accounting of related work — a related-work section in final-report-ready form — along with an updated plan for finishing something by the end of the semester.
For teams of two, it should contain an explanation of who did what so far and the planned split of work for the remainder of the project.
The final report should be at least eight pages, written in LaTeX. It is a detailed, conference-submission-style presentation of the entire research project, and should be written as a conference submission: abstract, introduction, related work, background, the details of your novel contribution, methodology, results, supporting evidence, and a conclusion with future work.
For teams of two, it should contain a section describing each person's individual contribution.
The presentation is a five-minute summary of your main research findings. Get across the cool ideas, and show a demo or a cool bug or attack that you found.
For students who are unfamiliar with the research process, especially producing a research paper, the following resources will be useful:
The purpose of the presentation is not simply to summarize the assigned paper. Everyone in the seminar is expected to have read the paper before class. Your job is to help the class understand the paper more deeply: explain the most important ideas, identify what is interesting or difficult about the work, critically evaluate its claims and assumptions, and lead a productive discussion.
A strong presentation should therefore devote substantial attention to analysis, interpretation, and discussion, not merely to reproducing the structure or contents of the paper.
Total: 100 points
The presentation demonstrates a strong understanding of the paper and explains its most important ideas accurately and clearly.
A strong presentation will:
Clearly identify the problem or research question the paper addresses.
Explain why the problem matters, particularly from a privacy perspective.
Provide the background necessary for the class to understand the contribution.
Clearly explain the paper's main ideas, methodology, system, model, or argument.
Identify important assumptions, including the threat model or adversarial model when relevant.
Explain the main results and what those results actually establish.
Select the technical details that are important for understanding the paper rather than attempting to reproduce every section.
Excellent (18–20): Demonstrates deep command of the paper. Explanations are accurate, selective, and insightful; difficult ideas are made substantially easier to understand.
Good (16–17): Demonstrates solid understanding and explains the major ideas accurately, with only minor omissions or areas lacking clarity.
Satisfactory (13–15): Correctly communicates the paper's basic contribution, but important details, assumptions, or results receive superficial treatment.
Needs improvement (0–12): Contains significant misunderstandings or omissions, or does not adequately explain the paper's central contribution.
The presentation goes beyond reporting what the authors say and offers thoughtful independent analysis.
Possible forms of analysis include:
Identifying important strengths and weaknesses.
Questioning assumptions or aspects of the threat model.
Evaluating whether the evidence supports the authors' claims.
Identifying missing experiments, comparisons, controls, arguments, or measurements.
Discussing alternative interpretations of the results.
Connecting the paper to other papers, techniques, or ideas from the course.
Considering how the work would behave under different deployment conditions.
Discussing how changes since the paper was published affect its conclusions.
Identifying promising extensions or open questions.
For privacy papers in particular, useful questions may include:
What notion of privacy is the paper attempting to provide?
Who is protected, and from whom?
Which adversaries or harms fall outside the model?
What information is assumed to be public or private?
Is the paper measuring privacy itself or a proxy for privacy?
What privacy/utility tradeoff is being made?
Who receives the benefit of the proposed system, and who bears its risks?
Are the users, datasets, or populations studied representative of the people affected?
Are there legal, social, organizational, or deployment assumptions that materially affect the conclusions?
Excellent (14–15): Provides substantial independent analysis and identifies important issues or connections that are not obvious from simply reading the paper.
Good (12–13): Provides meaningful critique and interpretation beyond the authors' presentation.
Satisfactory (10–11): Includes some analysis, but the presentation remains primarily a summary of the paper.
Needs improvement (0–9): Provides little evidence of independent evaluation or synthesis.
The presentation has a clear intellectual structure and makes effective use of the available time.
A strong presentation will:
Organize the material around the ideas needed to understand and evaluate the work rather than mechanically following the paper section by section.
Make clear which ideas and results are most important.
Use figures, examples, diagrams, or excerpts where they improve understanding.
Use readable slides that support rather than replace the spoken explanation.
Allocate time appropriately between background, explanation, analysis, and discussion.
Excellent (9–10): Exceptionally clear and well organized; the presentation makes the paper easier to understand.
Good (8): Generally clear and effective, with minor organizational or pacing issues.
Satisfactory (7): Understandable but occasionally difficult to follow, overly detailed, or poorly balanced.
Needs improvement (0–6): Organization, pacing, or presentation materials substantially interfere with understanding.
The presenter is prepared to explain the material and communicates it effectively.
This category concerns preparation and intelligibility, not performance style. You are not being graded on charisma, polish, or whether you appear nervous.
A strong presenter:
Is clearly prepared and familiar with the paper.
Explains ideas rather than simply reading slides.
Uses terminology precisely.
Keeps approximately to the allotted time.
Is able to clarify points when the audience is confused.
Excellent (5): Clearly well prepared and communicates the material effectively.
Good (4): Prepared and generally clear, with minor issues.
Satisfactory (3): Adequately prepared but noticeably dependent on notes/slides or occasionally unable to explain material clearly.
Needs improvement (0–2): Insufficient preparation significantly affects the presentation.
Presenters should come prepared with several substantive questions or issues for discussion. Good questions should invite analysis rather than simply test whether classmates remember facts from the paper. (Remember also that your class participation grade depends in part on thoughtfully answering questions posed by the presenter!)
Strong discussion questions might:
Challenge an assumption made by the authors.
Ask whether the experimental evidence supports a particular conclusion.
Ask whether a different threat model would change the result.
Compare the paper's approach with another technique.
Examine an apparent tradeoff.
Explore implications the authors did not consider.
Ask whether the proposed privacy goal is the right one.
Ask how the system might fail in realistic deployment.
Ask what experiment, theorem, study, or piece of evidence would most increase our confidence in the paper.
Invite disagreement about whether the paper ultimately succeeds.
During discussion, presenters should listen to what classmates say, follow up on interesting points, connect comments to one another, and allow the discussion to move beyond their prepared questions when productive.
Excellent (23–25): Initiates and sustains a substantive discussion. Questions expose important issues in the paper, and follow-up questions help the class develop ideas rather than merely collecting responses.
Good (20–22): Provides strong discussion prompts and successfully generates meaningful conversation, though some opportunities for deeper follow-up are missed.
Satisfactory (17–19): Comes prepared with reasonable questions and produces some useful discussion, but questions tend to be broad, obvious, or quickly exhausted.
Needs improvement (0–16): Discussion questions are superficial, factual, or poorly prepared, or the presenter makes little effort to develop class discussion.
This category evaluates the presenter's ability to engage with questions in real time. Presenters should be prepared to respond to questions about both the contents of the paper and their own analysis.
You are not expected to know the answer to every question. Saying “I don't know” is entirely appropriate when you do not know. Strong performance consists of understanding what is being asked, reasoning carefully from the paper and your own knowledge, distinguishing evidence from speculation, and engaging seriously with challenges or unexpected questions.
A strong presenter will:
Understand questions posed by classmates and instructors and respond to the substance of the question.
Answer questions about the paper accurately when the answer is established by the paper.
Demonstrate command of details that may not have appeared explicitly in the prepared presentation.
Distinguish what the authors actually demonstrate from conjecture, interpretation, or the presenter's own view.
Reason productively about hypothetical changes to assumptions, threat models, experimental settings, or system designs.
Connect questions to relevant concepts elsewhere in the paper or course.
Defend an interpretation or criticism when appropriate, while being willing to revise it in response to a convincing objection.
Recognize when a question exposes a limitation or issue that the presenter had not previously considered.
Acknowledge uncertainty rather than bluffing or giving an unsupported answer.
Excellent (23–25): Demonstrates deep command of the paper and responds thoughtfully to difficult, unexpected, or probing questions. Reasons effectively beyond material explicitly prepared in advance and engages substantively with challenges to their interpretation.
Good (20–22): Answers most questions accurately and thoughtfully and demonstrates a solid ability to reason about issues beyond the prepared material.
Satisfactory (17–19): Handles straightforward questions adequately but has difficulty with deeper questions, hypotheticals, or challenges requiring synthesis across different parts of the paper.
Needs improvement (0–16): Frequently cannot explain important aspects of the paper when questioned, gives unsupported or evasive answers, or demonstrates substantially less understanding during Q&A than suggested by the prepared presentation.
A presentation that gives an accurate and polished summary of the paper but offers little independent analysis or generates little meaningful discussion will not receive an excellent grade.
Likewise, the prepared slides and presentation are only half of the assignment. Presenters should expect to demonstrate their understanding of the paper through unscripted discussion and Q&A. Being able to explain, defend, reconsider, and reason about the work in response to questions is a central learning objective.
Presentation style is intentionally a relatively small component of the grade. A presenter who is nervous or not especially polished can earn an excellent grade by demonstrating deep understanding of the work, explaining the important ideas clearly, offering thoughtful analysis, and engaging seriously with the seminar discussion.
The central question to keep in mind when preparing is:
What can I do during my presentation and discussion that will leave the rest of the seminar understanding this paper—and the questions it raises—better than they did after reading it on their own?
Students are expected to attend and remain engaged during seminar meetings.
10%: Consistent attendance; at most one unexcused absence.
8–9%: Occasional absences or lateness.
5–7%: Repeated absences or lateness.
0–4%: Frequently absent.
Excused absences will not count against the attendance grade.
Students are expected to contribute regularly to seminar discussions. Strong participation includes asking thoughtful questions, responding substantively to classmates, connecting papers to other course material, and raising useful critiques or observations.
Quality matters more than quantity: students do not need to speak constantly, but they should contribute meaningfully and consistently over the course of the semester.
9–10%: Regularly makes thoughtful, substantive contributions and asks strong questions that advance discussion.
7–8%: Participates consistently and makes useful contributions, though usually without substantially advancing the discussion.
5–6%: Participates occasionally, but contributions are limited, superficial, or inconsistent.
0–4%: Rarely or never contributes to discussion.