Course Details
Course Description
DS 593 is about privacy-related challenges that computer systems face. The top-level question is whether we can design computer systems that better protect their users' privacy, while maintaining the desirable features, functionality, and convenience of today's applications. Some of the questions we will look at include:
- Where should a user's data be stored? On the user's own computer? In a datacenter run by a company like Google, Amazon, Facebook, or Microsoft? In commercial cloud storage services such as Amazon S3? In cooperative, decentralized peer-to-peer storage?
- What does current legislation require from the software systems that store and process our data? Are the current systems adequate?
- How do we translate abstract legislative requirements (e.g., the rights granted to users by the EU's GDPR) into technological solutions?
- What is the role of encrypted storage and data transmission? Can it help us preserve our privacy, given that many services we use need to process our data in various ways? Who should hold the keys?
- Can we specify precise privacy policies for our applications? Can we rely on systems to automatically enforce these policies?
- What are the right trust models for services that hold and process user data? How paranoid should we be?
- To what extent are the business models (e.g., targeted advertising) of current "free" web services incompatible with the desire for data privacy, and are the technologies that preserve both privacy and profitability?
- What performance overheads are acceptable for better privacy protection?
- How can we protect individuals' privacy while releasing meaningful aggregate statistics? How can we balance privacy and accuracy?
- Can we deeply embed user data privacy as a design concern in the computer systems we build? How disruptive would such a change be?
- How do we evaluate the usability of our tools, both in terms of developer effort and risk of misuse?
I (Kinan) don't know the answers to many of these questions, nor do I know if these questions are the right ones to ask; I hope to learn as much as you do from this course.
We will look at research papers on web services, datacenter systems, distributed communication systems, machine learning techniques, and privacy legislation. During the course, you will present and discuss papers, finish small hands-on assignments, work on a research project, and present your project at the end of the semester. There will be no exams.
Requirements
This is an advanced, research-focused course, but undergraduate students are most welcome! The course staff are keen to guide you, but the course will benefit us all the most if you take initiative. Students should know something about web or distributed systems and have some experience programming in systems languages before taking this course. For example, this might result from having taken one or more prior courses, such as DS 210, DS 310, CS 350, CS 451, ECE 413, or ECE 440. This is not an exhaustive list of how students may satisfy the course's prerequisites. Interested students that have not taken any of these courses are welcome, but should get in touch with Kinan before signing up.
Time Commitment
You can expect to spend approximately 3 hours per week in class (40 hours total), and around 3 hours preparing readings (40 hours total). Assignments and the final project will take upwards of 8-10 hours per week (112-140 hours total).
Grading
Your in-class presentations and participation, and your final project make up the bulk of your grade.
We will provide grades for specific assignments and are happy to discuss your academic standing on request. In general, if you've kept up with the material, and it is clear that you put significant effort into the class discussion, the paper presentations for which you were the lead and your project, you'll do well.
Grading breakdown:
- Reading Questions: 10%;
- Assignments: 15%;
- Participation: 15%;
- Presentations: 20%;
- Final Project: 40%.
Policies
Late Policy
Since this is a reading-centric discussion course, there are no formal provisions for late submission. We expect you to attend every session, but let Kinan know if you have any special requirements. For sickness and other issues of wellbeing, please obtain a note from health services and We will accommodate them.
This course will involve substantial reading for each meeting, and you will need to stay on top of the assigned readings to keep up, as we quickly move between topics. However, the research papers are relatively standalone, so finding one paper difficult to read will not disadvantage you going forward.
Reading questions are due at 11pm (Eastern) before the relevant session, and owing to the small amount of credit they contribute individually, there will be no late submission. If you do encounter particular, unexpected hardships however, send an email to Kinan.
Use of AI Tools
We do not allow any use of AI tools for the following:
- Preparing paper presentations.
- Any of the course assignments, including both their coding and their written portions.
- Writing the project proposal and the project report.
We allow the use of AI tools for the following, subject to the conditions below:
- Reading papers and comprehending the literature.
- Brainstorming ideas for project proposals.
- Coding assistance on the final course project.
When reading papers, we disallow using AI tools to provide summaries or answer our homework questions directly. Instead, if you want to use AI, use it as a thinking buddy to help you understand specific passages of the text. For example, by asking it to rephrase difficult sections, provide additional examples, or point to other related resources or background literature.
On the final course project, we disallow "whole-sale" use of AI tools, which compromises the learning process. For example, relying on a tool to generate the entirety (or large portions) of your code. Use such tools instead as thinking buddies for brainstorming, or as coding assistants.
In all cases, students remain responsible for their own understanding of the readings and for everything they submit. This includes being able to explain and answer questions about their code, its design and implementation, and their written work, and being able to participate effectively in discussion, both in class and asynchronously.
Students who use AI tools on the final course project must report that use in an appendix to their project report. The appendix should include:
- a log of the interactions between them and the AI tools (this can be link(s) to AI conversations).
- a description of which exact tools were used.
- an explanation of how the AI tools were used.
- a description of the reason for using the AI tools (e.g., save time, experiment with LLM, assist with understanding a specific API).
If we find that a student produced any of the disallowed work above with the help of AI tools, we reserve the right to withhold the entire grade for that work and/or to report the incident as a violation of academic integrity.
We refer students to the CDS Guidelines for AI in Coursework for more context on this policy.
PaperBuddy
At the BU SPACE lab, we are developing a novel, research AI assistant tuned to be a paper reading assistant or a "PaperBuddy". This assistant is meant to resemble a socratic tutor or a TA, it may help guide students in their readings, e.g., by asking them questions, pointing them to relevant pieces of the text, providing additional examples, or positioning the work within the context of broader literature.
We will deploy PaperBuddy in a limited, experimental way in this class. We will make PaperBuddy available to students for three of our readings. The students may elect to use it for these readings or to ignore it and do these readings as they would normally. We will use this experience to improve the course, PaperBuddy itself, and our understanding of how students use AI in flipped classroom settings.
PaperBuddy is built on top of Claude, a commercial large language model developed by Anthropic. By using PaperBuddy, students consent to their questions and their interactions with it being sent to Anthropic, where they are processed subject to Anthropic's terms of service and privacy policy. Students who prefer not to share their interactions with a third party should simply not use PaperBuddy and do the reading as they normally would; this carries no penalty of any kind. As with any online tool, students should not enter sensitive or personal information into PaperBuddy.
PaperBuddy also records students' interactions with it, such as the questions they ask and the responses it gives. These logs are used for research only for students who opt into the study described below; for everyone else, they are used solely to operate, maintain, and debug the tool. In no case are they used for grading: whether a student uses PaperBuddy, how much they use it, and what they say to it have no bearing on their grade, and the teaching staff will not look at any student's interaction logs until after the semester ends and final grades are released. Students with questions or concerns about any of this are welcome to raise them with Kinan at any time.
We will ask students to participate in an optional study about PaperBuddy and the use of AI in flipped classrooms. In this study, we will ask students short survey questions about their experience with PaperBuddy. We expect the survey to take a few minutes for each of the three readings. Participation is totally optional: students will not be penalized or rewarded for participating or not participating in the study. Furthermore, participation and students' survey answers will not be visible to the teaching staff until after the semester ends and the final grades for the course are released.
Accommodations
Boston University is committed to full inclusion of all students. Please inform Kinan if you have a disability or other condition that might require accommodations or modification of any of these course procedures. You may email Kinan, come to office hours, or speak with Kinan after class, and your confidentiality is respected. We will do whatever we can to support accommodations recommended by BU. For more information contact BU Disability Services at 617-353-3658, or access@bu.edu.
Mental Health
Being a student can be very stressful. If you feel you are under too much pressure or there are psychological issues that are keeping you from performing well at BU, we encourage you to contact BU’s Counseling Services. They provide confidential counseling and can provide notes supporting extensions on assignments for health reasons.
Incomplete Policy
We expect everyone to complete the course on time. However, we certainly understand that there may be factors beyond your control, such as health problems and family crises, that prevent you from finishing the course on time. If you feel you cannot complete the course on time, please discuss with Kinan the possibility of being given a grade of Incomplete for the course and setting a schedule for completing the course in the upcoming year.
Thanks to James Tompkin and Tom Doeppner for the text on accommodation, mental health, and incomplete policies.