Repository navigation
Data weekly scoreboard: commitments and forecasts for 2026-10-01 #1013
Description
Commitments and pre-registered forecasts from the 2026-09-24 data weekly, to be scored at the next one (Thursday 2026-10-01, 10:00 ET).
Each item has one owner, a yes/no resolution and a probability set on 09-24. At the start of the next meeting, mark each item yes or no, compute the Brier score, and carry any open item forward with a fresh forecast. Don't re-forecast mid-week; if something changes, note it below.
Decisions to settle by Friday 09-25 (Max)
- d178 provenance: build the certified candidate from scratch on one main commit (recommended).
- d179: drop the ty2020 W-2 tips return-count row at every vintage (recommended).
- d098 publish shape: certified non-default first, or straight to default. Proposed flip condition: flip when all terminal gates and both preflights are green, the QRF waiver (d177, approved 09-24) is bound in the manifest, and a smoke comparison against the current default shows no regressions beyond the waiver columns.
- Name owners for County-calibrated County Files (3,144 counties, 2026 + 2027): build epic #922 (County Files, with a first milestone date), publish_release 409s on the publish-non-default-then-flip-later flow (tag exists) #450 (the publish-then-flip 409) and State top-tail income mass is uncalibrated: CO M+ AGI at 24% of SOI in buildp-acs-local (fails progressive-bracket scoring) #940 (the state top-tail income mass).
Forecasts for 2026-10-01
| id | Owner | Resolves yes if | P |
|---|---|---|---|
| route-a-published | Max | A Route A certified release is published, with all terminal gates and both preflights green | 0.60 |
| route-a-default | Max | ...and it is the default dataset. Assumes d098 allows a direct flip; about 0.05 if it waits on #578 | 0.45 |
| remediation-merged | Max | The five Route A remediation PRs and #992 are merged to main through their gates | 0.80 |
| post-export-measured | Max | The post-export scoring stage (smoke plus reform validation) is measured at full scale, with a wall time | 0.90 |
| abawd-real-hours | Max, David | David's SNAP ABAWD analysis runs on an ACS-local file with real usual hours (depends on the SPM contract work) | 0.50 |
| uk-2024-25-default | María, Vahid | #966 is merged and the UK 2024-25 national default is promoted (needs Max's #823 answers by Friday) | 0.65 |
| uk-mortgage-tenure | María | Mortgage debt is measured by tenure, with a reply in Nikhil's 08-19 thread | 0.90 |
| uk-student-loan-gate | María | The #1006 student-loan gate is resolved | 0.80 |
| graph-green-and-plan | Max | #1011 is fully green, plus a written plan that separates the US graph switch from the ACS multi-spine work, with a date | 0.75 |
| ci-tests | Anthony, Ziming | #1009 is merged, plus a from-scratch test-suite proposal with baseline and target runner counts | 0.70 |
| diagnostics-subpackage | Anthony | #1007 is merged | 0.60 |
| dashboard-lookup-deploys | Anthony | The variable-lookup build cut and automatic dashboard deploys are live | 0.85 |
| dashboard-route-a | Anthony, Max | The calibration dashboard shows the Route A release with its gate evidence and the QRF waiver | 0.35 |
| issue-triage | All | Open microcosm issues are at 150 or fewer, and each remaining one has an owner or a later label |
0.50 |
Machine-readable
A plain ledger of this week's items, scored here in the issue. It is not a plangraph input: plangraph plans month-scale work and derives its schedule, while these are hand-set one-week forecasts. The month-scale microcosm roadmap (Route A through the flip, the UK default, the graph switch, County Files) is being modeled separately as a plangraph plan.
scoreboard: microcosm-data-weekly
setOn: 2026-09-24
resolveOn: 2026-10-01
seats: [max, david, maria, vahid, anthony, ziming, all]
items:
- {id: remediation-merged, lane: us-release, label: "Route A remediation PRs and #992 merged", owner: max, forecast: {p: 0.80}}
- {id: post-export-measured, lane: us-release, label: "Post-export scoring measured at full scale", owner: max, forecast: {p: 0.90}}
- {id: route-a-published, lane: us-release, label: "Route A certified release published", owner: max, predecessors: [{id: remediation-merged}, {id: post-export-measured}], forecast: {p: 0.60}}
- {id: route-a-default, lane: us-release, label: "Route A release is the default", owner: max, predecessors: [{id: route-a-published}], forecast: {p: 0.45, conditional: "d098 allows a direct flip"}}
- {id: abawd-real-hours, lane: us-local, label: "ABAWD analysis on real usual hours", owner: david, forecast: {p: 0.50}}
- {id: uk-mortgage-tenure, lane: uk, label: "Mortgage debt by tenure answered", owner: maria, forecast: {p: 0.90}}
- {id: uk-student-loan-gate, lane: uk, label: "#1006 student-loan gate resolved", owner: maria, forecast: {p: 0.80}}
- {id: uk-2024-25-default, lane: uk, label: "#966 merged and UK 2024-25 default promoted", owner: maria, predecessors: [{id: uk-mortgage-tenure}], forecast: {p: 0.65}}
- {id: graph-green-and-plan, lane: graph, label: "#1011 green plus a decoupled US graph-switch plan", owner: max, forecast: {p: 0.75}}
- {id: ci-tests, lane: infra, label: "#1009 merged plus a test-suite proposal", owner: anthony, forecast: {p: 0.70}}
- {id: diagnostics-subpackage, lane: dashboard, label: "#1007 merged", owner: anthony, forecast: {p: 0.60}}
- {id: dashboard-lookup-deploys, lane: dashboard, label: "Variable-lookup cut and auto deploys live", owner: anthony, forecast: {p: 0.85}}
- {id: dashboard-route-a, lane: dashboard, label: "Dashboard shows Route A with gate evidence", owner: anthony, predecessors: [{id: route-a-published}, {id: diagnostics-subpackage}], forecast: {p: 0.35}}
- {id: issue-triage, lane: backlog, label: "Open issues at 150 or fewer", all owned or labelled later, owner: all, forecast: {p: 0.50}}Posted from the 09-24 meeting thread in #microcosm.
🤖 Generated with Claude Code
Activity