research-document

Software Engineering Research Portfolio Queue

Portfolio Objective

Reduce the uncertainties that most affect whether the first bounded pilot can produce valid, safe, decision-relevant evidence at acceptable operating cost. This is a generated-by-review navigation view; mission and research records are the durable nodes.

Prioritization Method

Rank considers decision consequence, uncertainty, dependency leverage, falsifiability, feasibility, and risk. Rank is provisional and must be revisited after every completed item. A blocked high-value item does not displace useful unblocked work.

Active and Backlog Graph

Rank Node Maturity Parents Dependencies Enables Confidence Largest uncertainty State
1 MS-PREREG-2026-6B9F Deep Investigation HY-HANDOFF-2026-6B31 preflight, ethics, accessibility, statistics, independent review EX-HANDOFF-2026-0D7A Medium design sample size and external gates active; external-dependency
2 EX-HANDOFF-2026-6E42 Validation measurement, safeguards, baseline independent reviewers, accessibility participants, ethics determination confirmatory experiment Low rubric reliability and protocol feasibility proposed; external-dependency
3 MS-REVIEW-2026-7A31 Validation four review REPs and three decisions independent reviewer acceptance or revision Low independent survivability external-dependency
4 EX-AICODING-2026-2D77 Scoping HY-AICODING-2026-7C2D partner, privacy-reviewed telemetry, preregistration AI-use policy Medium theory current causal moderators blocked
5 MS-ROSATTRIB-2026-31D9 Idea ROS pilot first handoff milestone ROS adoption decision Low credible project counterfactual proposed; dependency-blocked
6 MS-TERMS-2026-0F4D Idea DF-BOUNDARY-2026-3F72 none label disposition Very Low any positive label value proposed; low expected value

Completed Nodes Still Awaiting Review

Node Maturity Finding Review debt
MS-SAFEGUARDS-2026-89D1 Synthesis minimal-data sandboxed study is feasible in design ethics, data steward, and participatory accessibility gates
MS-BASELINE-2026-E5A7 Synthesis conventional note is artifact control; ROS needs separate study comparator plausibility and decision review
RP-MEASURES-2026-71B4 Synthesis safe continuation plus separate harm/cost gates rubric reliability and independent review
RP-AICODING-2026-B7E2 Synthesis AI effects are conditional independent source/disposition review
RP-HANDOFF-2026-A8E1 Synthesis handoff evaluator is best provisional slice validate outcome model and baseline
RP-BOUNDARY-2026-5A7C Synthesis distinct discipline unsupported independent boundary review
DF-HANDOFF-2026-5C9B Synthesis select reversible handoff pilot remains review
DF-BASELINE-2026-48C0 Synthesis select competent conventional-note control remains review
DF-BOUNDARY-2026-3F72 Synthesis reject distinct discipline remains review

Knowledge Relationships

RP-HANDOFF / HY-HANDOFF
  -> MS-MEASURES (complete) ───┐
  -> MS-SAFEGUARDS (complete) ─┼-> EX-PREFLIGHT
  -> MS-BASELINE (complete) ───┘      -> MS-PREREG -> EX-HANDOFF
                                                         |
first handoff milestone ---------------------------------┴-> MS-ROSATTRIB

RP-AICODING / HY-AICODING
  -> partner + safeguards + preregistration -> EX-AICODING

RP-BOUNDARY / DF-BOUNDARY
  -> MS-TERMS (low priority)
  -> comparative profile test embedded in EX-HANDOFF

Monitoring and Review Triggers

  • Re-rank after every mission completion or material contradictory evidence.
  • Review AI-coding evidence when tool generation changes materially or by 2026-10-30, whichever comes first.
  • Review handoff/profile conclusions after the first pilot evidence.
  • Move any item backward if construct validity, safety, or replication weakens.
  • Never advance an item to Engineering Ready while its governing Decision Record remains in review.

Portfolio Review — 2026-07-30

  • Priority change: Outcome validity was resolved enough to stop blocking design. Safeguards and baseline are now complete in synthesis.
  • Invalidation: The earlier queue ordering was invalidated; it placed baseline and safeguards ahead of the measurement dependency and treated reconstruction time as if already usable.
  • Merge/archive decision: No duplicate or obsolete mission was found. Historical completed work remains visible because it carries review debt.
  • Repository decision: No new repository is justified. All active questions concern this pilot and share its evidence graph.
  • Highest-value opportunity: Determine whether the study can collect enough evidence while minimizing source exposure, behavioral surveillance, accessibility exclusion, and participant risk.

Portfolio Review — after safeguards and baseline

  • Priority change: Preregistration is the highest-value active item, but its next steps require independent reviewers, accessibility participants, qualified ethics determination, data stewardship, and statistical review.
  • Invalidated assumptions: An identical time cap is not accessibility neutral; de-identification alone is not a privacy boundary; a no-documentation control is not credible; a handoff artifact experiment cannot validate ROS.
  • Queue split: Whole-ROS attribution moved to MS-ROSATTRIB-2026-31D9. Rubric/protocol feasibility moved to EX-HANDOFF-2026-6E42 so it cannot be confused with confirmatory efficacy.
  • Feasibility challenge: A simple scenario needs about 388 recipients per arm for a 50% versus 60% success contrast at 80% power and two-sided 5% alpha. The bounded pilot is therefore preflight research unless a justified larger effect, stronger design, or adequate sample is established.
  • Archive/merge decision: No record should be archived. The original flat queue is superseded as a navigation model but preserved as a compact view.
  • Highest-value next action: Obtain the external roles and approvals needed for EX-HANDOFF-2026-6E42; otherwise independent review has higher expected value than low-priority terminology work.