A practical guide for Mentors and the IPMC

This guide summarises governance activity on incubator-general from 2015 to 2025 by analysing:

• subject lines from incubator-general mboxes
• podling population data from podlings.xml
• trends in voting, reporting, discussion, and release-related threads

Participation is measured only by the number of unique email addresses per thread.

Summary

This guide explains what data from 2015 to 2025 shows about Incubator governance activity on the incubator-general list. It combines:

• governance thread data from incubator-general
• podling population data from podlings.xml
• vote duration, and participation patterns

The main findings are:

• governance activity on incubator-general rises and falls with podling population
• most governance threads on incubator-general are release votes
• release frequency per podling has increased over time
• vote duration is a useful indicator of overall system load
• participation diversity shows how visible and distributed governance discussions are
• the Incubator is most predictable when active podlings remain under about fifty
• silence, lack of releases, and persistently low participation diversity are strong early warnings

This guide helps mentors and the IPMC interpret the list activity based on measured data rather than anecdote.

1. The Incubator Behaves Like a Population, Not a Queue

Governance activity closely follows podling population levels. Higher podling counts correspond to more governance threads, more votes, and more total traffic.
Lower counts reduce overall traffic, although governance threads make up a larger share of the list.

This reflects demographic changes, not shifts in expectations or culture.

2. Governance Is Mostly Release Work

Across 2015 to 2025 there were:

• 3,958 governance threads
• 2,919 are VOTE threads (~74%)

Most are release votes, with fewer graduation or retirement votes.

Release work exposes licensing, NOTICE, and packaging issues and is the clearest indicator of real progress.
Note: release vote counts include both successful and failed attempts (multiple release candidates).

3. Release Activity, Fewer Podlings but More Frequent Releases

When release votes per year are divided by the number of active podlings, we see:

• 2015–2017: ~2.2 to 2.8 releases per podling
• 2018–2022: ~3.1 to 3.7
• 2023–2025: ~4.3 to 5.0

Podlings are releasing more frequently relative to their number.
They tend to reach their first release sooner and maintain a steadier cadence.
The Incubator is smaller, but podlings appear more active in release work.

4. Release Vote Reliability

Repeated release votes with the same version number provide a proxy for revotes.
Most releases pass on the first attempt, but some versions appear multiple times in a single year. These repeated attempts decline sharply after 2020, even adjusting for a lower podling population.
This suggests improved release hygiene, clearer documentation, and better mentoring support.

5. Graduation Patterns in the Data

Graduation vote counts vary by year. The clearest influence is intake timing:

• high intake years are followed 1–2 years later by more graduations
• lower intake years are followed by fewer graduations

Graduation timing depends on community maturity, release quality, and sustained engagement, not chronological cycles.

6. Retirements Follow a Two to Three Year Lag on Average

Retirements were more frequent early in the dataset, with a notable spike in 2023.
Lack of releases and list silence around month twelve remains the clearest early sign of trouble.
The data shows timing, not causes, so interpretation should remain cautious.

7. Vote Latency and Podling Population

Median vote duration per year varies from 7 hours to 37 hours. Median vote duration is always under 72 hours. Faster years (2022–2025) align with lower podling counts and indicate high reviewer availability.
Higher podling populations tend to correlate with more variable vote durations.
Vote latency is a helpful indicator of system load, though not a precise function of podling count.

8. Observed Capacity of the Incubator

Patterns across 2015 to 2025 suggest that the fastest behaviour appears only once podling counts fall to the low forties. Above 60 podlings, vote durations are never that fast.
These are heuristic observations, not hard limits, but they suggest that maintaining podling counts below about 50 supports faster, more predictable voting behaviour.

9. Release Cadence and Retirement Risk

The data does not provide causation but shows a strong association:

• podlings with no releases in the first year are far more likely to retire later
• podlings that release early tend to sustain participation and move toward graduation
• rising release cadence corresponds with lower long-term retirement rates in later years

Faster releases appear to reduce retirement risk, or at least correlate strongly with podlings that continue to build community. This aligns with long-standing mentoring experience and is supported by the observed patterns.

10. Governance Behaviour on incubator-general

10.1 Governance Is Becoming More Focused

Earlier years show higher overall message volume.
Later years show lower traffic but a higher share of governance threads (10–13 percent in most years after 2017).
The list carries less incidental noise and a higher proportion of structured governance activity.

10.2 Participation Diversity

Threads vary from low diversity (1–2 participants) to high diversity (6 or more).
Higher diversity indicates broad engagement; lower diversity indicates concentrated participation.
Consistently low diversity in a podling’s governance threads may signal concentrated governance and warrant mentor attention.

10.3 Thread-Level Participation and Message Concentration

Most governance threads involve small groups and short discussions:

• the largest share of threads include only 1–2 participants
• most threads contain fewer than 15 messages

Longer or larger threads are uncommon and typically relate to graduations, retirements, naming issues, or problematic release candidates.
These patterns are stable and suggest predictable governance behaviour.

10.4 Ratio of DISCUSS to VOTE Threads

DISCUSS threads are always far fewer than VOTE threads, reflecting that:

• release votes rarely require a DISCUSS thread
• DISCUSS is used mainly for graduation, retirement, trademark, or process questions

DISCUSS counts decline after 2017, while VOTE counts remain stable.
Governance discussions appear more focused and predictable over time.

11. What Mentors Should Take Away

1. Early slow growth is common.
2. Release cadence is the strongest visible health indicator.
3. Silence and lack of releases after the first year correlate with later retirement.
4. Vote latency reflects overall reviewer load.
5. Participation diversity highlights whether governance is shared or concentrated.

12. What the IPMC Should Take Away

1. Governance load follows podling population trends.
2. Vote latency is a practical indicator of capacity.
3. Release frequency per podling has improved over time.
4. Participation diversity reveals oversight patterns.
5. Monitoring release cadence, silence, and low diversity is more informative than raw message counts.

Methodology and Limitations

Data Sources

The analysis uses:

• 10 years of Incubator general mailing list mboxes
• podlings.xml

Approach

1. Governance thread detection based on subject lines.
2. Vote duration from thread start and end timestamps.
3. Participation diversity from unique email addresses per thread.
4. Podling population derived from startdate and enddate.
5. Governance share of traffic from governance vs other messages.

Shortcomings

1. Subject-only classification can misclassify unusual subjects.
2. Scope limited to incubator-general, not podling dev lists or private lists.

Use of AI

AI was used to help normalise subjects, summarise data, and draft narrative sections.
All numeric values come directly from the generated CSV files and podlings.xml, and all statements have been reviewed to ensure they align with the data.

Final Summary

From 2015 to 2025, incubator-general displays consistent governance behaviour. Governance activity follows podling population, release work dominates governance threads, release cadence per podling has increased, and vote durations provide a useful indicator of overall load.

Participation, diversity, and silence remain strong early-warning signals. The Incubator operates across a wide range of podling counts, but the data suggests that lower podling counts support faster and more predictable governance behaviour.

These findings help mentors and the IPMC interpret list activity using measured data rather than anecdote.