Project rubric

Back to the course homepage

The innovative, open-topic project contributes 30% of the course grade, split across four milestones:

Projects are assessed against their approved question, method, and evidence. There is no project competition or common leaderboard. For logistics, see the project milestone guide.

1. Project pitch (10 points)

Criterion Points Description
Privacy problem and stakes 2 The pitch identifies a concrete user, system, or scientific setting, protected asset, and meaningful privacy problem.
Innovation and baseline 2 The closest existing system or method is clear, along with the distinct capability, attack, defense, measurement, or insight the project may add.
Technical direction 2 The proposed artifact or research method is concrete enough to evaluate for coherence and course fit.
Evidence and feasibility 2 The pitch names a plausible evaluation path, available resources, and a serious scope or technical risk.
Delivery and response 2 The presentation is focused, uses its time effectively, and answers a brief question accurately.

2. Proposal (5 points)

Criterion Points Full-credit standard
Problem and privacy setting 1 States the problem being solved and the relevant threat model or privacy property.
Innovation and baseline 1 Identifies a cited baseline and explains the specific contribution beyond it.
Technical approach and progress 1 Describes the proposed method and provides a current artifact, experiment, or proof component.
Evidence and evaluation 1 Presents an initial result or documented failure and a plan to evaluate the main claim with a baseline and metric or formal target.
Execution plan 1 Gives a timeline with responsibilities and identifies a main risk with a fallback plan.

Scoring: Each row has two main requirements. Award 1 when both are provided and supported, 0.5 when only one is, and 0 when neither is. For partial or zero credit, identify the missing or unsupported part. A documented failure can earn full evidence credit.

3. Poster/demo (10 points)

Criterion Points Description
Contribution and technical substance 3 The completed contribution, method, and system or research artifact are understandable and technically sound.
Innovation 2 The presentation demonstrates a meaningful difference from the closest baseline or prior workflow.
Evidence 2 The presentation shows concrete results, comparisons, failures, or analytical support rather than only describing an idea.
Poster or demonstration quality 2 The poster or live demo makes the workflow and main result easy to inspect and is readable or reliable in the presentation setting.
Limitations and response 1 The team answers questions accurately, states limitations without overclaiming, and identifies each member’s contribution.

4. Final report (5 points)

Criterion Points Description
Technical execution and contribution 2 The report explains a correctly executed method and the completed contribution relative to the baseline.
Evidence and baselines 1 Results use appropriate comparisons and auditable evidence at a scale suitable for the claim.
Analysis and limitations 1 Conclusions match the evidence and address failures, residual risks, and threats to validity.
Clarity, artifacts, and contributions 1 The report is organized and cited, and supporting artifacts and team contributions are documented appropriately.

Additional notes

  1. Innovation is required, but publication-level novelty is not. A project must add a meaningful capability, attack, defense, measurement, system boundary, interaction, or insight. A reproduction, routine comparison, literature review, mechanical port, or incremental extension alone does not satisfy the requirement.
  2. Claims should match evidence. A narrow, carefully tested conclusion is stronger than a broad claim with thin support.
  3. Auditability matters. Include counts, splits, seeds, prompts, configs, examples, derivations, or logs as appropriate to the project.
  4. Different formats need different evidence. A system needs tests and realistic workflows; an attack or defense needs a threat model and controls; a monitor or auditor needs verified positives and clean cases; an empirical paper needs a defensible design; and a theoretical paper needs precise assumptions and reasoning.
  5. Visuals support communication. Figures and tables should make the main comparison or conclusion easy to see, but polish cannot replace evidence.
  6. Individual and group scope may differ. Individuals may submit narrower projects. Teams of 2 should show broader experiments, stronger comparisons, or a more complete system.
  7. Contribution statements are required. Substantial contribution imbalances may lead to adjusted individual grades.