Open Contribution — Current Bet #2: every idea gets a qualified response

[lfnovo]

Decision (Sep 5, 2026): modify — the qualified response stays; a cadence commitment is added: every new Idea gets its qualified response within 3 working days, with the facilitation loop running daily. Full data and reasoning in the decision comment. Next review: October 6, 2026.

Bet #1 separated exploration (Discussions) from execution (Issues). We kept it — routing worked. But it exposed the next gap: ideas arrived well-formed and then sat there. Most Ideas Discussions had zero replies. Routing alone doesn't create deliberation; it only creates the space for it.

Our current bet

If every idea gets a qualified response — decomposed into its distinct needs, verified against the actual code, given an explicit status, and routed to where each part belongs — then conversations start, and the path from idea to shipped work becomes visible.

A qualified response is not "thanks, we'll consider it." It means:

• Decomposition: most ideas contain several distinct needs (one recent "thumbnails" idea contained three; "manual annotations" contained four). Each need can have a different answer.
• Verification: claims get checked against the code before we reply. This week that check found two real bugs the reporters didn't know they'd found.
• Explicit status: every idea ends with one of — exploring (real problem, open solution space, with questions that shape the design), accepted → graduated (small and well-defined: becomes a work item immediately), parked until champion (valid, needs a community owner), combine (folded into a canonical Discussion), or answer.
• Routing: recurring themes get one canonical Discussion instead of five scattered threads; fixes that belong in our upstream libraries (content-core, esperanto) get tracked there and linked.

Transparency on how: this facilitation is AI-assisted, exactly as described in The Bigger Picture — AI prepares the analysis and verification; every outcome and every reply is reviewed and approved by a maintainer before posting. The maintainer remains the decision owner; AI makes the depth sustainable.

Starting point

The bet starts with the backlog cleared. As of today:

• all 21 open Ideas Discussions have a qualified response (10 from the new intake form, 11 pre-form);
• 5 work items were born from them, each with acceptance criteria (Search UI/docs don't say what text search covers (insights are searchable but undocumented) #1267, Notes markdown editor hardcodes light mode, ignoring the app theme #1268, Support CRAWL4AI_API_TOKEN for remote Crawl4AI instances (content-core bump + docs) #1269, plus content-core#80 and esperanto#276 upstream);
• 2 real bugs were found purely by verifying claims against the code;
• 2 canonical Discussions were established: packaging channels and multi-user — the latter with an open call for scenarios;
• 3 stale multi-user Issues were consolidated (two closed as evidence, the umbrella kept).

What we will observe

Over the next four weeks:

• Time to qualified response for new ideas — does it stay days, not weeks?
• Deliberation: do authors answer the sharpening questions? Do third parties join?
• Graduation quality: do the Issues born here get implemented without rework?
• Calls to action: do the multi-user scenarios (Multi-user support — what should it mean for Open Notebook? #1254) and channel champions (Packaging & deployment channels — TrueNAS, Unraid, Helm, and how they get supported #1250) materialize?
• Sustainability: does the maintainer-approval loop keep up with intake volume?

What would change our mind

If qualified responses don't generate follow-up engagement — authors don't return, questions go unanswered, canonical Discussions stay silent — then depth of response isn't the bottleneck and we'll look elsewhere. If the approval loop can't keep pace with intake, we'll rescope before quality drops.

On September 15, 2026 we record the decision: keep, modify, or revert — and what we learned.

[lfnovo]

Decision: keep the qualified response, fix the cadence — modify

Ten days early, because the data is in and the queue is empty. Here is what four weeks showed against each criterion we said we'd watch.

We said we'd observe	What happened (Aug 16 → Sep 5)
Time to qualified response	The Aug 16 cohort: same day. The three ideas that arrived afterwards — #1276 (Aug 18), #1281 (Aug 23), #1283 (Aug 25) — waited 15, 10 and 8 days. All were answered on Sep 2.
Deliberation	6 authors came back to answer the sharpening questions (#765, #1215, #1236, #1243, #1254 twice, #1265) and 1 third party joined (#1276). Before this bet, replies on Ideas were near zero.
Graduation quality	Of the 5 work items born from the queue, 4 are implemented and closed (#1267, #1268, #1269, #1299) with no rework; the upstream fix lfnovo/content-core#80 shipped in content-core 2.0.7. Verifying claims against the code found 3 real bugs the reporters didn't know they'd found.
Calls to action	#1254 (multi-user): one complete scenario, refined over two rounds. #1250 (packaging channels): no champions yet.
Sustainability	The approval loop stalled for two weeks in the second half of August. Nothing was left unanswered in the end, but the latency was entirely ours.

Reading: the depth of the response is not the bottleneck — it created conversations, produced work items that turned into code without rework, and converted one call to action. What it exposed is the next bottleneck: cadence. A qualified response that arrives two weeks late is still a two-week silence.

What changes

The qualified response stays the standard. What's added is a cadence with a public commitment:

From today (Sep 5, 2026), every new Idea gets its qualified response within 3 working days.

How: the facilitation loop now runs daily — the analysis, verification against the code, precedent search and draft replies are prepared every day, and the maintainer reviews and approves before anything is posted (same rule as before: AI prepares, the maintainer decides). The daily rhythm is what makes 3 days a promise instead of a hope.

Also adopted along the way, and now part of the process: a new status, incubating — direction decided, timing open, the Discussion stays the home — and the rule that work items are born when someone is going to build them, not when the idea became clear. ready is a promise of execution; we don't want it to become a graveyard.

What we will observe next

Until October 6, 2026:

• The 3-day commitment: hit rate, measured on every Idea opened from today.
• Deliberation: does the faster turnaround keep authors in the conversation? Do third parties keep joining?
• Incubating ideas: do they get built when their time comes, or do they quietly die in the Discussion?
• Packaging & deployment channels — TrueNAS, Unraid, Helm, and how they get supported #1250 champions: does anyone step up, or does "parked until champion" need a different mechanism?
• The vision thread (Evidence-Centered Research — where Open Notebook is heading #1315): does a written direction change the quality of what arrives in Ideas?

What would change our mind

If the daily loop can't sustain 3 days, we'll say so and reset the number rather than let it slip silently. If faster responses don't increase deliberation, speed isn't what the community needed and we'll look at the questions we ask instead.

On October 6, 2026 we record the next decision: keep, modify, or revert.