Joshua Poddoku

Joshua Poddoku

@joshuapoddoku.com

Building Governance for Organizations & Communities | Open Source Evangelist 🌐 joshuapoddoku.com Views my own.

83 followers219 following108 posts

Pinned

Every thread about AI and open source ends in blame. AI-generated PRs nobody reads. AI slop shipped as code. Maintainers get the burnout either way. I think it's shared responsibility. Both sides have to understand each other for the human part to survive.

Tech is finally healing. Building software is shifting back to pure FOSS, giving me 90s vibes. The era of corporate "Open Source" harvesters extracting value for pure greed is ending.

They pressure developers into providing free engineering under the guise of "building social proof." When the software breaks? They blame "low-quality public contributions." It’s corporate exploitation dressed up as community.

Governance used to sound like corporate jargon. Now projects and contributors need it. A pull request alone no longer shows real involvement. Projects need people. People need projects structured well enough for their contributions to matter. What have you seen work?

Projects need contributors who bring: - Context - Ownership - Thoughtful review Contributors need: - Clear expectations - Support - A path in - Room to learn and grow Both lists are now non-negotiable to keep standards.

Protect your human agency at all costs. We need to take care of AI stewardship (now SI), as we need to govern it as a resource rather than a tool.

I’d like to connect with grant programs, investors or VCs on how they evaluate applications based on their AI policy. Great questions to start with, if you haven’t asked them yet, what LLMs did you use, for what intent, what were the prompts, and how did that help you articulate.

Ever heard of "technological solutionism"? It’s the fallacy that every complex human problem has a technical fix. When we prioritize the "how" of the technology over the "why" of the user's pain, we stop solving real problems and start building toys. We need solutions that serve the community.

Kudos to Tuta for advocating for privacy! This commitment helps the international community stand up for digital rights and privacy. I've personally DeGoogled most of my life -- highly recommend making the switch. #Privacy #DigitalRights #DeGoogle #Tuta

Tuta@tuta.com

🚨🇪🇺 The #ChatControl trilogue starts on Tuesday & the EU Council is trying to pressure Parliament into giving up its opposition. We must stop them once again! 💪🏼 Because no matter the framing, the truth remains the same: There's no backdoor for the good guys only! ➡️ fightchatcontrol.eu

Open Source is real when the person who starts a project sparks a ripple of innovation among its contributors. A new contributor should do better than the last. That’s how you measure whether a project can thrive.

Joshua Poddoku reposted

Since Jan 1, I've received 200+ vulnerability reports and reviewed every one. Most looked AI-generated, and some days different people reported the exact same issue. I assigned just four CVEs. Over 98% were AI slop or unverified. This is the daily reality of OSS.

Every thread about AI and open source ends in blame. AI-generated PRs nobody reads. AI slop shipped as code. Maintainers get the burnout either way. I think it's shared responsibility. Both sides have to understand each other for the human part to survive.

I built this framework to keep human-AI collaboration healthy. Five areas keep a project alive: - Brain: your thinking, the one thing you bring - Community: contributor experience - Content: the DevRel work - Code: good first issues, docs - Governance: what AI touches, what stays human

what matters the most is your "community" Not the one you built the "corporate way" who come in, get the help, and exit. We need more of natural communities if want to build a thriving OSS tech business.

It's simple to understand why community comes first before code Code is the vehicle but community is the driver. The driver's health and wellbeing comes first before everything existed. Code can fail. But community won't if it is built with right principles and intent.

This is proudly powered by @astro.build and will be sharing real-time updates on the state of AI-written code.

Joshua Poddoku@joshuapoddoku.com

59% of 1,799 posts about AI-written code sound wary of AI. Mood: Pushback, 19/100. Community Signals by @joshuapoddoku.com www.joshuapoddoku.com/community-si...

1/ How do Open Source projects stay open when AI pull requests take hours to review? Since Sep 4 I've been tracking 1,700+ posts on AI-generated code. These are the problems that keep coming up 🧵

2/ • Low-effort AI PRs eating maintainer review time • Projects closing the door, like libvips and Forgejo • Unclear copyright on AI-written code • Vibe coding and careful AI-assisted work treated as the same thing

Joshua Poddoku reposted

I'd say if you're a longterm thinker, then the answer should be "Community". Community can give birth to new beginnings, get people built on trust and even hire them in the future. This keeps things happening naturally and healthy for businesses. Businesses can fail historically.

Just published a working paper: The Shop That Closes at Five. ​It details Clansurf project that studies small, community-governed nodes connecting local value providers and patrons without internet, apps, middleman fees, or platform tax. ​Read it here: www.clansurf.com

clansurf.comThe shop that closes at fiveLocal providers are invisible to the people nearby. A working paper on Clansurf: offline, community-governed nodes that help neighbours find each other.
View full thread

4/5 It is automated, it cannot read sarcasm, and it is not a survey. Read it as a direction. Every quote links to the original post. I send a short note on what it shows every two weeks. You can sign up on the page.

2/5 Here on Bluesky, 63% of posts react to AI code without getting into it. On Hacker News, 60% are hands-on: building, debugging, reviewing. HN also reads more positive, 55% against 39% here.

3/5 Here's the news for maintainers. libvips now refuses any contribution made even partly with generative AI. Forgejo also bars AI from commit messages, PR descriptions and review. Turso takes the other side: use AI, but understand and test what you send. (Source @cmdpromptinc.bsky.social)

Joshua Poddoku reposted

I keep hearing folks say "delete all your skills" when a new model comes out. No - skills can be pipelines of work, not just "this is how you interact with a service". Audit, eval, don't just blindly delete. I wrote about this here: jimbobbennett.dev/blogs/skills...

jimbobbennett.devDon't delete your skills, audit themBoris Cherny says to delete your CLAUDE.md, skills, and hooks every six months. That's good advice for one kind of skill and dangerous for another. Here's how to tell them apart, and why the skills th...