Release Vote Insights

A Decade of Lessons from Apache Incubator Release Votes (2014–2025)

Introduction

This report summarises ten years of Apache Incubator release-vote activity to identify long-term patterns in community behaviour, process efficiency, and mentoring outcomes. It documents how the ASF’s culture of openness and collaboration translates into measurable improvement over time. The goal is transparency and shared learning, rather than evaluating individual communities or mentors.

Executive Summary

Over ten years of Apache Incubator release votes, it has evolved from rule enforcement to a culture of mentoring and shared accountability. Analysis of about 33,000 emails across 1,600 vote threads shows a clear pattern: fewer rejections, faster collaboration, and a stronger, more confident generation of ASF communities.

Key changes

• Fewer rejections: −1 votes fell from roughly one in six in 2016 to fewer than one in ten by 2025.
• Better first attempts: Most releases now pass on the first release candidate (RC).
• Tone shifted: Conversations moved from “You must fix this” to “Let’s fix this together.”
• Policies improved: Clearer documentation steadily reduced procedural mistakes.
• Regular cadence: Podlings that release frequently and maintain engaged mentors are consistently associated with successful graduation outcomes.

The Incubator’s history demonstrates that mentoring, openness, and consensus are effective practices, showing how ASF culture strengthens through open collaboration.

1. Overview and Context

This analysis was conducted to understand how Incubator release practices have evolved and to identify any measurable signals of community health or mentorship engagement. It covers all public general@incubator.apache.org release-vote threads between January 2015 and September 2025, spanning votes on approximately 160 Podlings, including both now-graduated and retired projects.

During that period, approximately 9,500 emails, 1,600 vote threads, and 1,000 distinct individuals participated in release votes. A rotating core of 15–20 active IPMC members accounted for two-thirds of all votes at any given time. This core was supported by a broader ring of 200–240 mentors who participated in release-vote discussions over the decade, averaging about 45 active mentors per year. Participation rotated gradually, with about one-third of mentors active in 2018 still active in 2024, showing healthy renewal and continuity rather than stagnation.

2. Cultural and Process Evolution

2. Cultural and Process Evolution (2014–2025)

2014–2017 – Policy-Driven

• Typical tone: “This violates ASF policy.”
• Release votes were compliance-oriented and defensive.
• Mentors focused on enforcement and procedural accuracy.
• Threads were long and detailed, with frequent references to policy sections.
• This era established necessary governance discipline after rapid Incubator growth.

2018–2020 – Standardised

• Typical tone: “Please check …” — firm but constructive.
• Release checklists, vote templates, and reproducible builds became routine.
• Clearer policies (Distribution Policy, WIP Disclaimer) reduced friction.
• Re-votes declined as podlings adopted consistent release practices.
• The IPMC began to operate more predictably, with fewer corrective interventions.

2021–2025 – Collaborative

• Typical tone: “Let’s fix this before the next RC.”
• Votes focused on guidance and education rather than enforcement.
• −1 votes became teaching opportunities, often paired with suggested fixes.
• Mentors and podlings worked together to improve quality before formal votes.
• The process matured into a shared learning loop — less bureaucracy, more trust.

3. Mentorship and Continuity

About 200–240 mentors participated directly in release-vote discussions over the decade, with a core group of 20–25 carrying the sustained oversight load.

• Average active mentors per year: about 45
• Median tenure: 2.4 years
• Typical Podlings per mentor: 3–4
• One-third of 2018 mentors remained active in 2024, showing healthy renewal

Since 2020, clearer expectations for mentor engagement, along with improved onboarding and best-practice resources, have strengthened both participation and retention.

4. Release Cadence and Review Efficiency

Steady Cadence with Shorter Cycles

Podlings maintained a consistent release rhythm, averaging 3–4 releases per year once stable. The median release interval shortened from about 60 days (2016–2018) to 40–45 days (2022–2025), reflecting growing confidence, improved CI/CD, and better automation. More mature podlings often sustained monthly or bi-monthly release candidates, demonstrating operational readiness for graduation.

Reduced Review Overhead

Release-vote threads became shorter and more focused over time.

• Early on, discussions averaged 12–15 emails per thread, often running for a week or longer.
• By 2024–2025, they averaged 5–7 emails, with a median vote duration under 4 days.
  Efficiency improved thanks to clearer vote templates, standardised release policies, and stronger mentor pre-vote guidance.

Higher First-Pass Success Rate

The proportion of releases approved on the first vote attempt rose from about 65% in 2016 to over 80% by 2024. Early −1 votes were most often caused by missing or incorrect LICENSE and NOTICE attributions, unbuildable source packages, or artifacts staged outside official distribution areas. By the early 2020s, these mechanical issues had declined sharply as checklists, policies, and clearer documentation became standard practice. Automation and shared mentor checklists virtually eliminated recurring blockers such as bundled binaries, signature errors, or checksum mismatches. Today, most remaining −1 votes involve legal or provenance-related concerns, issues that require human judgment rather than mechanical validation.

Cultural Shift in Reviews

Mentors shifted from procedural enforcement to collaborative quality assurance. By 2021, most −1 votes became educational blockers, offering feedback to improve the next RC rather than outright rejection. Podlings increasingly pre-validated artifacts before calling a vote, turning reviews into confirmation rather than discovery.

5. Policy Reforms and Their Effects

Each reform was followed by fewer recurring mistakes and smoother votes, suggesting that clearer documentation supports more consistent practice.

6. Release Rhythm and Graduation

Correlation Between Cadence and Success

Podlings that released regularly (every 1–3 months) were far more likely to graduate within 2–3 years, while those with long gaps between releases often struggled with community participation or mentor engagement.
By the early 2020s, release cadence had become a practical indicator of project health and readiness for graduation.

Evolution Over Time

• 2014–2017: Release schedules were irregular and mentor-driven; many podlings struggled with compliance.
• 2018–2020: Checklists, templates, and automated standardised expectations helped podlings establish predictable release rhythms.
• 2021–2025: Regular cadence became the norm, with most mature podlings maintaining two or more releases per year.

Observed Patterns

• Podlings with at least two releases per year typically graduate within 24–30 months.
• Podlings with 6+ month gaps were overrepresented among retired or inactive communities.
• Projects maintaining a steady cadence showed stronger contributor growth and mentor confidence.
• Many long-running podlings that delayed graduation did so despite technical maturity, often because of low community velocity rather than release-quality issues.

7. Governance and Values in Practice

Release oversight remains anchored in peer-to-peer, community-driven governance. Policy guidance flows laterally through mentors and community consensus rather than through formal intervention.

8. Early-Warning Indicators for Struggling Podlings

Tracking these indicators helps the IPMC and mentors intervene early to address community, process, or communication issues before momentum is lost.

9. Methodology

Source: general@incubator.apache.org archives (Jan 2015 - Sept 2025).
Processing: Automated detection of release-vote threads and −1 issue classification.
Scope: The analysis covers only public release-vote threads on general@incubator.apache.org. Private mentoring, off-list collaboration or podlings dev lists is outside this dataset.
Limitations: Private mentoring is not visible; mentor-training effects are still emerging.
Threads were identified by subject lines containing [VOTE] and [RESULT] for the same release candidate.