Replan: team formed, topic chosen, schedule computed
What was asked
Two weeks after the kickoff the facts had changed: a team of five, a topic, a proposal draft rendered by a teammate, an official submission repo from the professor, and a workshop meeting this week. The ask was to bring the tracker up to date and, from an engineering-management point of view, make the timeline honest: a real critical path rather than a hand-drawn one, a Gantt chart, float, and a cleaner site.
What was built
Data. data/activities.csv was rewritten as a 54-row work breakdown with planned dates, finish-to-start dependencies, hard deadlines on the submission activities, status and percent complete. deadlines.csv and weeks.csv were corrected against the schedule page (HW2 Sep 8, HW3 Sep 18) and marked up to today. data/risks.csv is new: ten risks with likelihood, impact, trigger, response and the activities that carry the response.
Engine. R/helpers.R gained cpm(): a forward pass from today as the status date (planned starts as start-no-earlier-than constraints), a backward pass from the official deadlines, total and free float, critical and near-critical flags, and an overdue flag. schedule_health() compares duration-weighted earned progress with what the plan says should be done by today. embed_schedule() writes the computed schedule into the page as JSON.
Views. assets/tracker.html is a dependency-free script, inlined into every page, that draws the Gantt (early-schedule bars with progress, late windows, deadline diamonds, a today line, the course-load lane, filters, hover details, click-to-trace dependencies) and the deliverable chain with the float on each handoff. New pages: Schedule and Risks. The home page leads with the next deadline, schedule health and the zero-float count. The critical-path page is now computed, not typed. The team page carries the roster, the topic, the attributes table and the ownership grid (owners TBD until the next sync).
Proposal review. The Sep 15 draft was read against the 55-point rubric. Blocking fixes: an AI citation artifact and a chat-tool tracking parameter left in the text, a doubled “Figure Figure 1” cross-reference, heading levels that break the template’s structure, and no references at all. Improvements: separate the buyer (developer or operator) from the respondent (renter); name the Prolific pipeline; replace the archetype competitor table with real products and prices; quantify the retrofit market; fix the zero-dollar price level and anchor the premium to a base rent; define bundled levels in one line each; say how research question three is answered. Fit risk: the emerging and pre-commercial criteria, to be argued at the workshop. Both statistics in the draft were verified against their sources.
How it was verified
The site was rendered end to end with Quarto and R and every page was opened in a browser at laptop and phone widths; the Gantt filters, the phase selector, the tooltips and the click-to-trace were exercised by hand, and the browser console was clean (a jQuery error from kableExtra’s print helper was found and removed by dropping that package). The CPM was prototyped independently in Python on the same CSV and the floats matched the R output row for row. The proposal chain shows zero float this week (workshop and deadline in the same week), the final-survey chain one day by design, and the fielding-to-report link zero. The public-site rule was checked: first names only, no handles, no credentials, no grades.
Same day, second pass: scope check against the course
Every deliverable page on the course site was re-read and its rubric copied into data/requirements.csv (33 criteria, 270 points); the new Requirements page maps each criterion to the activities that produce it and rolls their status up. Gaps it found and fixed: the design parameters (alternatives per question, questions per respondent, the no-choice option) were nowhere in the plan (now in SP03 and PS02); the eligibility text has to sit inside survey.qmd (FS03); the market simulation needs a defined competitor set (new FA07); the pilot report needs an example-question screenshot (PS05). The context bank’s “2 to 4 attributes” line was wrong (the course sets no count) and was corrected. The proposal draft was read again line by line and is now reflected on a project page: product statement, buyer versus respondent, the six attributes with metrics and open points, the three research questions with how each is answered, the market opportunity as the draft states it, the draft’s five open questions, a fit check against the seven selection criteria, a twelve-entry decision log (data/decisions.csv) and a timeline of the direction so far (data/timeline.csv). One real mismatch surfaced: the draft names the GWU community and personal networks as the respondent pool while the course default is Prolific with an eligibility pre-screen; it is decision D07 and risk R11, to be settled in the survey plan. Tooltips now open on tap on phones.
What is next
Book the workshop slot, revise the proposal to v2 (the review checklist sits in the team’s brainstorming repo under 1-proposal/), move it into the official repo and push by Sunday (P05 to P08). Get the fifth member added to the official repo and confirm the team size (S05). Assign owners at the next sync (S06) and decide who hosts the survey (S07). From Sep 21 the survey plan chain starts with zero float.