Strong work in this course

Every point in this course is published in advance in the rubrics on the four assignment pages. Read backwards, those rubrics show what the highest-weight rows reward, how the participation math works, and where the built-in flexibility is, so none of it depends on guessing what the instructor wants.

Strong work on the Ethics Case (25%)

The Ethics Case rubric puts 65 of 100 points on three rows: sociotechnical analysis (25), ethical frameworks (20), and mitigation (20). In practice that means:

  • One named decision behind the outcome, with primary-source evidence. The 25-point row asks where the harm or the success entered, which means a decision someone made and someone could reverse. “The AI was biased” does not name one; “auto-adjudication was switched on with no human review” names one.
  • Work each NASW value against a specific part of the system. The 20-point frameworks row scores values that do work in the argument. A list of five values pasted at the end scores as decoration.
  • Assign every mitigation an owner. The 20-point mitigation row pays for a mitigation someone could be held to, so “more oversight” earns nothing where “the county board requires an annual disparate-impact audit before contract renewal” earns the row.
  • Verify everything an AI told you against a primary source. The evidence row is 15 points, and some case packets contain one planted false fact that AI assistants will repeat, so the row is largely scored on whether you caught it.

Strong work on the Digital Story (25%)

The Digital Story rubric weights objective and audience (25), ethical representation (20), and craft (20) highest:

  • Make one ask of one audience. The 25-point row wants a named segment, a behavioral ask specific enough to name a time and a place, and a reason that audience should believe the ask will work. Restraint scores here: a video that follows one person, uses one number and makes one ask scores above one that piles on three statistics and a montage, and a piece that ends on “learn more” leaves the row empty.
  • Answer the standing question before you are asked it. The 20-point row starts with your positionality paragraph and the ethics standard you named and applied. Consent is a production step with a name and a date: anyone real who appears watches the cut and holds a veto, and the reflection says what changed after they saw it.
  • Build accessibility at the storyboard, and submit the caption file. Corrected captions, a transcript, a contrast check, and alt text are worth 15 points. The line that is checked is whether a corrected .srt or .vtt is in the submission, not whether captions appear when you press play.
  • Visible influence from the video notes. The 20-point craft row grades production quality against your form’s own conventions and whether your numbers come from a named source, and the 10-point video-notes row checks whether the videos you noted shaped what you made.

Strong work on the Strategy Pitch (25%)

The Strategy Pitch rubric puts its biggest single weight (20) on the Week 11 capacity brief, then 15 each on ethics and the briefing:

  • Do the Week 11 brief properly. It is one page and the largest single row, and it is the only part of this assignment you can finish early. It wants the four-part assessment, the interview done or booked with an actual date, the question in one sentence, and an honest account of what an hour could not establish.
  • Run the protocol on your own proposal and retain the honest section. The 15-point ethics row wants the top risk named with its mitigation and the conditions under which you would walk away. A well-argued recommendation against adoption scores full marks.
  • Make the briefing survive a listener who has not read the memo. The 15-point row is scored on whether someone in the room can repeat your recommendation and your top risk afterward.
  • Write evaluation and accessibility for the decision-maker. Concrete commitments and measures a reader can check earn 10 points each, and a statistics plan written for a methodologist earns neither.
  • Budget the true cost. Staff time is usually the biggest line, and the budget row wants the sustainability line (who owns this in year five) and a priced lower-cost alternative.

Participation math (25%)

Participation is three weighted pieces: attendance and discussion (8%), the Problem Inventory (10%), and the weekly question board (7%). Design Walls are bonus.

The inventory needs five entries across the semester, at least two by Week 6 and at least two more after Week 9, once the interview with your organization is underway. Entries are graded on quality, not volume: a specific scene, a specific who, an honest account of why the problem persists, and reasoned tech relevance. Design Walls are optional and worth up to three bonus points, and a post needs one sentence naming a design decision and who it includes or excludes. The weekly question board counts your best eight graded answers, so a missed week costs nothing.

Flex days and the other slack

The late-work policy has the flex days and what happens after they are exhausted, and Problem Inventory entries take no late credit, because their deadlines are the Week 6 and Week 9-window checkpoints themselves. Two bonus points are available for coming to office hours early in the term and booking a second visit to discuss a major assignment, and three more are available for event reflections.

Study habits

Verify AI claims against primary sources. The habit is graded directly on every assignment: the Ethics Case evidence row, the story reflection’s AI field, and the strategy pitch’s verified vendor and cost claims. An unchecked claim that proves wrong costs points in that row, and the planted facts in the case packets are there to test the habit.

Name a specific decision. “The algorithm did it” earns nothing anywhere in this course. A timeout setting, a data-matching choice, an auto-adjudication switch, or a lending policy is the kind of decision the biggest rubric row on the Ethics Case, the graded sentence on every Design Wall post, and the honest slide in the pitch all ask for, because a person made it and a person can reverse it.

Run the protocol on successes too. Option B cases are graded to the identical standard plus the conditions-of-success questions, the pitch runs the protocol on your own proposal, and Side B of the protocol asks whether the intervention would replicate without its champions.