The Death of Community Memory  ## Metadata - Author: [[JA Westenberg]] - Full Title: The Death of Community Memory - Category: #articles - Document Note: --- tags: [community, knowledge-management, communication] author: JA Westenberg source: Discourse Blog reference: [https://blog.discourse.org/2025/11/the-death-of-community-memory/] date created: 2025-11-14 00:00 --- # 🏷️ Title: The Death of Community Memory ## 🧭 Table of Contents - [📝 Summary](#-summary) - [📖 Main Content](#-main-content) - [🗒️ Notes](#️-notes) - [📚 References](#-references) - [🗣️ Paraphrase](#️-paraphrase) - [📰 Newsletter](#-newsletter) - [✅ Action Items](#-action-items) - [🧪 Hypotheses](#-hypotheses) - [📊 Methodology](#-methodology) - [❓ Questions](#-questions) - [🔮 Future Research](#-future-research) - [💡 Implications](#-implications) - [🧠 Learning Styles Assessment](#-learning-styles-assessment) - [📊 Accessibility Assessment](#-accessibility-assessment) - [🌍 DEIB Assessment](#-deib-assessment) - [✅ Logical Validity](#-logical-validity) - [🛡️ DARVO Analysis](#-darvo-analysis) - [🎯 Conclusion](#-conclusion) ### 🔗 References - Include links cited in or relevant to the article. ## 📝 Summary - What happened: The author describes losing critical decisions in ephemeral chat (“Chat is architecturally designed to be forgotten”) and argues for persistent, searchable forums/wikis for durable knowledge. - Why it matters: Organizations “get dumber as they grow” without accessible memory. - Key stakeholders: Product/engineering teams, community managers, OSS maintainers, support. - Risks and tradeoffs: Speed of chat vs. permanence; cultural inertia; invisible costs of rework. - Bottom line: Use chat for coordination; move decisions/explanations to persistent systems with clear norms. ## 📖 Main Content - Core arguments: - Real-time chat (Slack/Discord/Teams) is great for quick coordination but poor for memory and retrieval. - Forums/wikis provide searchable, structured, durable records (“Threads had descriptive titles… Search actually worked”). - Past ~20–50 people, loss of institutional memory accelerates without persistence. - Invisible costs: rehashing, repeating mistakes, slower onboarding. - Context/definitions: - Community memory: collectively accessible, persistent records of decisions, debates, tutorials. - Persistence media: forums, wikis, decision logs, post-call writeups. - Uncertainties: - Exact headcount thresholds vary by org; implementation success depends on culture and norms. ### 🗒️ Notes - “Chat is architecturally designed to be forgotten.” - Forums enable findability: titles, categories, threading, search. - Critical info to persist: technical decisions, policies, onboarding, rationales. - Growth pain: at 20–50 people, memory fragments; wikis without norms fail. - Invisible cost center: time wasted, avoidable mistakes, onboarding drag. - Prescriptions: chat for ephemeral; forums/wikis for decisions; move threads; write up after calls. ## 📚 References - Slack (https://slack.com) - Discord (https://discord.com) - Microsoft Teams (https://www.microsoft.com/microsoft-teams) - Discourse forum software (https://www.discourse.org) - Knowledge management basics (https://en.wikipedia.org/wiki/Knowledge_management) ## 🗣️ Paraphrase - Paragraph A: The author keeps losing important decisions in scattered chat and can’t retrieve them when needed. - Paragraph B: Forums once solved this with structure and search; we should pair chat (ephemeral) with persistent tools for decisions. ## 📰 Newsletter - This week: Why your Slack isn’t your memory. JA Westenberg argues that “chat is architecturally designed to be forgotten,” urging teams to pair real-time chat with persistent forums/wikis for decisions, policies, and onboarding. Actionable takeaway: create a decision log template, define “move-to-forum” triggers, and require post-call summaries. Tools to try: Discourse for threaded debates, a lightweight ADR workflow in your repo, and a “link-or-reply” norm in support channels. Bonus: audit one recent incident—how many minutes were lost to rehashing? ## ✅ Action Items #### DevRel - Publish “Chat vs. Forum” policy; demo ADRs; track resolution time and duplicate question rates. - Run workshops on writing durable decision summaries; office hours to migrate key threads. - Curate examples/templates: ADR, RFC, decision log; quickstarts for forum categories. - Campaign: “Link don’t retype”; attribute via forum link analytics. - Community ops: escalation from chat to forum; enforce code of conduct around findability. #### OSPO - Governance: define decision rights; require ADRs for architectural changes. - Compliance: add license/scanning docs to persistent knowledge base. - Security: establish vuln intake and embargo playbooks in a forum/Wiki. - SBOM: owners + freshness SLOs stored persistently. - Developer support: consultation hours; templated legal guidance. - Engagement: publish upstream agreements; track contributions in a visible ledger. - Strategy: OKRs tied to time-to-answer and duplication reduction. #### Community - Maintainers: roadmap clinics; release notes hygiene; governance FAQs. - Contributors: onboarding paths; first-issue rotations; pairing sessions. - Sponsors: transparent budgets and milestone reports in forum. - Ecosystem: cross-project incident drills; postmortems posted persistently. #### WordPress - Core: issue triage rituals; release criteria in forum; performance budgets. - Core Performance: profiling plans; metrics dashboards; perf bounties summaries. - Core AI: evaluation sets; responsible AI guardrails; decision logs. - Core Program Team: rituals, dependencies, retrospectives archived. - Mobile/Playground: repro labs; telemetry reviews; education modules. - Community/Training/Marketing/Support: meetup kits, lesson plans, editorial calendars, knowledge base gaps. - Design/Docs: UX audits; examples-first patterns; style guide in one place. - Photos/TV/Polyglots: moderation guides; caption quality; glossary alignment. - Accessibility/Meta/CLI/Hosting/Tide/Openverse/Themes/Plugins: publish criteria, checklists, sprints, provenance checks with findable links. #### The WPCC - Programs: fellowships and stipends with transparent criteria. - Projects: governance clinics; doc sprints; rotating release teams. - Funding: matched grants; milestone payouts; open ledgers. ### FAIR - Technical independence: signed artifacts; provenance. - Policies: funding transparency; sunset clauses. - Governance artifacts: reusable charters, operator agreements, scorecards. ## 🧪 Hypotheses - Teams that move decisions to forums reduce duplicate questions ≥30%. - Publishing ADRs decreases time-to-onboard engineers by ≥25%. - Post-call writeups cut rework from forgotten decisions by ≥20%. - Clear “move-to-forum” norms improve search success rates by ≥40%. ### 📊 Methodology - A/B across teams: chat-only vs. chat+forum; measure duplicate question rate, mean time-to-answer, onboarding time, incident rework. - Log analysis: forum linkbacks in chat; search queries vs. successful retrievals. - Surveys: perceived findability, cognitive load, decision clarity. ### ❓ Questions - What automated triggers should move chat to forum (message length, keywords, participants)? - Which structures (ADR, RFC, Q&A) maximize retrieval? - How to visualize “knowledge rot” to make costs visible? ### 🔮 Future Research - LLM-assisted summarization to seed forum threads from chat. - Decision provenance graphs across tools. - Incentive models for contribution to persistent knowledge. ### 💡 Implications - Practice: mandate ADRs and post-call summaries. - Policy: make findability a KPI. - Security: persistent incident runbooks reduce risk. - Education: structured learning paths in forum. - Funding: sponsor maintenance of community memory. ### 🧠 Learning Styles Assessment - Reading level: Flesch-Kincaid Grade ~8–9. #### 👀 Visual learners - Present: clear contrasts of tools. - Improve: flowchart of “when to move to forum.” #### 🔊 Aural learners - Improve: 3-minute audio summary. #### 🗣️ Verbal learners - Present: analogies, quotes. - Improve: glossary (ADR, RFC, decision log). #### 🧩 Kinesthetic learners - Improve: step-by-step “migrate a chat to ADR.” #### 🧠 Logical learners - Present: causal chain (chat → loss → cost). - Improve: decisio - Summary: I recently spent forty minutes searching through Slack trying to find a technical decision — made eight months ago — about an app that 60% of my creative life depends on...I eventually gave up and just asked someone to explain it again, which they did, poorly, because they didn’t quite remember either. - URL: https://blog.discourse.org/2025/11/the-death-of-community-memory/ The death of community memory a photo of Joan Westenberg I recently spent forty minutes searching through Slack trying to find a technical decision — made eight months ago — about an app that 60% of my creative life depends on. I knew the conversation had happened. I remembered participating in it. I could picture the rough time frame and vaguely recall who else was involved. None of this helped. The discussion had taken place across three different channels, and was then moved “offline” by the team, and the rest of us (The Community) had no idea where any of it had landed. I eventually gave up and just asked someone to explain it again, which they did, poorly, because they didn’t quite remember either. This keeps happening. Communities are having the same debates over and over. New members ask questions that were definitely answered six months ago. Teams rediscover solutions to problems they already solved. Users search for solutions to problems that seem to repeat. And repeat. And repeat. We used to have forums. And forums had one massive advantage: you could find things. Threads had descriptive titles. There were categories. Search actually worked because the content was structured for retrieval. If someone asked a question that had been answered before, you could link them to the previous discussion instead of retyping everything. Then Slack happened, and Discord, and Teams, and we all decided that real-time chat was simply better: More modern // more collaborative. More like how humans “naturally communicate” (as if there’s anything natural about the internet itself.) The real-time feedback loop is genuinely useful. Being able to quickly ask questions and get immediate answers removes a lot of friction. The problem is that we replaced forums entirely instead of using both tools for what they’re good at. Chat is architecturally designed to be forgotten. Messages flow by in streams and channels multiply until nobody knows where anything is. Important decisions get made in casual conversations that disappear into the archive. Search sort of works if you remember the exact words someone used, but good luck finding that nuanced technical discussion from last quarter when you only remember the general topic. I’m not even arguing that forums are better than chat for most daily communication. Chat is great for quick questions, coordination, social connection, all the small interactions that keep a team functioning. The issue is that we’re using chat for things that need to be permanent: Technical decisions, policy discussions, onboarding documentation, explanations of why we built something a particular way etc. All of this goes into Slack channels where it effectively disappears after a few weeks. The result is that organizations are getting dumber as they grow. Why? *Because community memory no longer exists in any accessible form.* Everything lives in someone’s head until that person leaves or forgets. Knowledge that was carefully built up and shared gets lost because it was shared in the wrong medium. Small groups can get away with this. When you’re five people, everyone is part of every important conversation. You don’t need an archive because the archive is your collective memory. But past a certain size, maybe around 20 or 30 people, you start losing things. Decisions get made that other people don’t know about. New members can’t get up to speed because there’s no coherent place to learn how things work. A community hits 50 people and suddenly nobody can find anything. They start talking about “documentation” and “knowledge management systems” and maybe someone sets up a wiki that nobody uses. The real solution would be to stop having important conversations in chat, but that feels impossible because the entire culture is built around Slack now… So they keep using chat and keep losing knowledge and keep wondering why everything takes so long. But this is a problem we know how to solve! The technology exists. Forums still work. Searchable, threaded, persistent discussions are not some lost ancient art. We’re choosing not to use them because real-time must be better, right? Because real-time communication feels more dynamic and exciting. And apparently, communities have to spark an adrenaline rush now? Companies pay for Slack per user per month. The cost of storage is real but abstracted. Meanwhile, the cost of fragmenting and decaying knowledge is completely invisible until it’s too late. How do you measure the time wasted rehashing old decisions? How do you quantify the mistakes that could have been avoided if someone had been able to find that old discussion? These costs are real and large, but they don’t show up in any budget line. I think there’s an easy answer in treating *different* types of communication *differently*. Chat for ephemeral coordination; forums or wikis or something persistent for decisions and explanations that need to be findable later. Clear norms about when to move a discussion from chat to somewhere permanent. Actually writing things up after important video calls instead of letting them evaporate. I’m tired of searching through Slack for answers that I know exist somewhere but can’t find. I’m tired of watching good ideas get lost because they were shared in the wrong channel. I’m tired of organizations that are smart individually but stupid collectively because they’ve destroyed their own ability to remember.​​​​​​​​​​​​​​​​ When I first got online in 2003, my home on the internet was a collection of video game modding forums—places like DieHard Wolfers, a Wolfenstein 3D community where people shared custom levels and argued about game design. I was probably too young to have anything interesting to say, but I posted anyway. Tips about texture mapping. Questions about making better enemies. Half-baked ideas about what made levels fun. Twenty-two years later, that forum is still there. The posts are still there. I can type my old username into a search box and find conversations I barely remember having. I can see how I thought about problems back then, what I was excited about, what I was trying to learn. The community's entire history is intact and searchable — every tutorial, every debate, every solved problem still available for anyone who needs it. Those forums were designed to be permanent and findable. And because of that, they still serve as the actual memory of a community. New modders can search for answers instead of asking the same questions. Experienced people can link to old discussions instead of rewriting explanations. That's what we've really lost: the idea that some conversations should be designed to last, that communities and companies need memory to function, that real-time and permanent aren't the same thing, and that *we need both.* I'd give a lot to have that kind of searchable permanence for the technical decisions we're making right now. To know that in 2047, someone could search for why we built something a particular way and actually find the answer. To build a community — or a company — that gets smarter instead of constantly forgetting itself. We know how to do this. We just have to remember that we want to. ##### Sign up for our newsletter! Get community tips and product updates delivered to your inbox We'll use your email to send communications about the launch and future marketing materials from us. Don't worry, we won't share your email with anyone else and you can opt-out anytime by clicking "unsubscribe"., The Death of Community Memory, JA Westenberg, 01k9zn692yrr1559vrd8vfttzf.