# uist-workflow

> Use when planning a UIST cycle end to end — the late-March abstract and paper deadlines, the ten-week build-freeze before video production, the rebuttal window, conditional acceptance, the July camera-ready, adjunct-track parallel lanes, and the November conference — with owners and risk buffers.

- Skill: `brycewang-stanford/uist-workflow` (Agent Skill)
- Install (CLI): `npx skillmds add brycewang-stanford/uist-workflow`
- Raw SKILL.md: https://api.skillmd.com/api/skills/brycewang-stanford/uist-workflow/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Productivity
- Author: brycewang-stanford (https://skillmd.com/u/brycewang-stanford)
- Updated: 2026-08-19
- Page: https://skillmd.com/skills/brycewang-stanford/uist-workflow

---


# UIST Workflow

A UIST year is engineered backwards from two fixed points: the paper deadline in late
March and the conference in the fall. Between them sits a pipeline with almost no
slack for hardware failures, video production, or study scheduling. This skill turns
the cycle into a managed project. All dates below are the verified 2026 numbers
(checked 2026-07-08); reopen `uist.acm.org/<year>/cfp/` before committing a calendar.

## The 2026 reference calendar

| Date (2026) | Event | Author obligation |
|---|---|---|
| Mar 24 | Abstract deadline (AoE) | Title, abstract, authors, track metadata in PCS |
| Mar 31 | Paper deadline (AoE) | PDF + video figure + supplementary, all anonymous |
| May 28 - Jun 5 | Rebuttal window | 5,000-character self-contained response |
| Jun 27 | Notification | Decisions are conditional on rebuttal-promised changes |
| Jul 24 | Camera-ready | TAPS source, alt text, final video, +10% page allowance |
| Nov 2 | Doctoral Symposium (in person) | Separate track, separate deadline |
| Nov 2-5 | Conference, Detroit | Talk + (typically) live demo of the system |

As of 2026-07-08 the live obligations are the camera-ready chain for accepted papers
and the adjunct tracks (Demos, Posters — deadlines 待核实, read the CFP) for everyone
else; new submissions target the 2027 cycle.

## Backwards plan for a systems paper

Work in reverse from the paper deadline; the ordering below is where UIST projects
actually fail, not a generic writing schedule:

1. **T minus 0-1 weeks — submission mechanics.** PCS uploads, anonymity sweeps,
   abstract-field consistency (see `uist-submission`).
2. **T minus 1-3 weeks — video figure production.** Scripting, capture, edit,
   captions, anonymization. The video cannot be shot until the system is stable, and
   it competes with the writing for the same people (see `uist-supplementary`).
3. **T minus 3-6 weeks — evaluation.** Technical benchmarks and any user study; IRB
   approval must predate this window, so file it in the fall.
4. **T minus 6-10 weeks — feature freeze.** Everything demonstrated in the paper
   must exist by here. Features added after freeze belong to the next paper.
5. **T minus 10+ weeks — build.** Include one full hardware re-spin or one major
   refactor in the estimate, because there will be one.

## Owners and risk register

Assign names, not roles, and write the mitigation before the risk fires:

```text
RISK                              OWNER      MITIGATION
Hardware breaks in video week     <name>     Build two units; film B-roll early
Study participants no-show        <name>     Over-recruit 30%; pilot in freeze week
Video render/caption pipeline     <name>     Dry-run export at T-3w with dummy edit
Page-limit overflow (10pp body)   <name>     Appendix plan agreed at T-4w
Rebuttal week hits a holiday      <name>     Pre-draft objection playbook (uist-author-response)
Camera-ready owner on vacation    <name>     TAPS + alt-text buddy named at notification
```

## The two parallel lanes

UIST is one conference with two archival containers, and a healthy group uses both:

- **Papers lane** (main proceedings): the full pipeline above.
- **Adjunct lane** (Demos, Posters — separate chairs, separate deadlines, adjunct
  proceedings): a rejected paper's system can still appear at the conference as a
  demo; early-stage work can test community reaction as a poster; the same work must
  not be submitted to both Demos and Posters (the 2026 rule converts it to a demo and
  drops the poster). Decide the fallback lane on notification day, not after the
  adjunct deadlines pass.

## Planning UIST 2027 from a mid-2026 vantage

For teams reading this in the live window (post-notification 2026), the next Papers
deadline is roughly nine months out and the calendar is not yet published (待核实 —
watch the 2027 site from its first announcement). Provisional planning rules that
survive cycle drift:

- Assume the abstract/paper pair lands in **late March ± a few weeks** and the
  conference in the fall; anchor the feature-freeze ten weeks before the assumed
  deadline and adjust when the real dates post.
- Anything involving **human subjects** needs ethics approval filed by roughly
  December; anything involving **custom hardware** needs its final revision ordered
  by January (fab and shipping queues do not respect AoE).
- Book **video-production capacity** — the person, the rig, the edit suite — for
  the two pre-deadline weeks now; it is the least substitutable resource in the
  pipeline.
- Draft the **contribution list before the spring**: if you cannot write three
  capability claims in January, the project is a 2028 paper wearing a 2027 badge.

## The month-by-month cadence (generic UIST year)

| Months before deadline | Focus |
|---|---|
| 9-7 | Routing decision, system architecture, ethics filing |
| 7-4 | Core build; related-work sweep running in parallel |
| 4-2.5 | Feature freeze; evaluation harness and pilots |
| 2.5-1 | Studies and technical characterization; paper skeleton |
| 1-0.5 | Video production; full drafts circulating |
| 0.5-0 | Submission mechanics only — nothing new enters the paper |

## Between cycles

- File a post-mortem within two weeks of notification while reviews are fresh:
  which review criticism was predictable at freeze time?
- Convert the submission's infrastructure — build scripts, benchmark harnesses,
  capture rigs — into reusable lab property; UIST groups compound or burn out on
  exactly this.
- Watch the next edition's site from its first announcement: dates shift by weeks
  between editions and the length rules have changed across recent cycles.

## Output format

```text
[Cycle target] UIST <year> papers / demos / posters / DS
[Today's phase] build / freeze / evaluation / video / submission / review / rebuttal / camera-ready / conference prep
[Critical path] <the single task that moves the finish date>
[Risk register] top three, each with named owner
[Next fixed date] <date + obligation>
```