We're building public alternatives to private AI knowledge.
Why We Exist
Stack Overflow's traffic dropped 78% in two years. Wikipedia g...
For further actions, you may consider blocking this person and/or reporting abuse
Sounds like a great initiative, really curious to see where it will be going - will you be posting updates on progress?
It sounds like a gargantuan task, but maybe I'm misunderstanding what you're trying to achieve (and how) - are you planning to build some sort of "vault" (comparable to the genome seeds vault which they've build somewhere in the arctic), or is it not like that?
Or else, what's the strategy gonna be?
Shouldn't you (also) try to stand with and promote existing sites and communities which DO still support "the commons" when it comes to (developer) knowledge - like SO, Wikipedia, Reddit, dev.to etc etc?
I'm just blathering a bit coz personally I don't really have any clear thoughts on this ...
we're not archiving old knowledge. we're preventing new knowledge from disappearing.
you solve something with claude → close tab → lost forever. next dev solves it again privately.
we built: save chats, auto-scan for secrets, share the good ones. federated (like mastodon).
updates coming soon 👍
Makes sense! Integrate knowledge preservation/archiving into the "AI process" - something similar crossed my mind, but you've got the vision and looks like you've got an idea of how to do it ... good stuff!
Yeah maybe referring to (and 'promoting') those sites/communities could be part of your mission - but, the idea to "hook into AI coding tools/processes" (Claude Code, etc) to archive knowledge from AI coding sessions, and make that publicly available (if I understood that correctly) - that sounds like a pretty brilliant strategy - I like it!
Love this iniative although I don't fully understand it yet.
It is often said now that we're loosing a generation of developers now. I am where I am because people trusted me, gave me enough time to finish my work and because I had to dig through problems the hard way without magic shortcuts.
youve named exactly what we're addressing
"losing a generation" is real. juniors get answers without seeing the reasoning, dead ends, or verification that made you strong.
the "hard way" taught: reading docs, debugging systematically, recognizing
patterns, understanding tradeoffs.
AI shortcuts skip all that. working code but no judgment.
what we're preserving: that reasoning publicly
not "heres the answer" but "heres how i verified this, what i tried first, why that failed, the tradeoff im making".
juniors need to see developers THINK, not just solutions
does that clarify?
That makes sense, thank you. I guess I can do my part and add more of the thinking process into my articles. This is real effort. Hats off to anyone investing this time.
Great Org! I would love to join you guys, specifically on the role of "Preserving developer knowledge"! Can't wait to see what you guys' post!
hey Francis! appreciate the enthusiasm.
we're keeping it small initially while we figure out what works, but youre welcome to contribute in comments and discussions.
whats your take on preserving developer knowledge? what techniques do you use for verification in the AI era?
if you have thoughts worth writing about, wed love to see them.
Hey Daniel! Sorry for the late reply. I tend to not be active on Weekends.
For your questions. I think having the fundamentals as a developer is critical, regardless to how technology change overtime. For example, knowing best practices when it comes to programming and knowing how to communicate with user's needs a building off with it.
The most important part is the interview process. Without AI, we would be practicing leetcode problems and practicing with peers. With AI, same process but we will ask the most efficient questions to practice. Makes up a lot of time spent and efficiently studying because we know what we need to work on.
That is my take on preserving developer knowledge, which is knowing the basic fundamentals of programming in both the developer and in the business setting as well as preparing for technical interviews. What do you think?
CloudFlare blocks me from accessing many sites, it blocks Grok from loading documents attached to a space, and acts as one of the biggest walls on the internet right now even above paywalls.
I would be concerned if today's gatekeepers become tomorrow's librarians
this is excellent pushback and exactly the kind of scrutiny we need.
youre right to be concerned about cloudflare as infrastructure. the irony isnt lost on me.preserving public knowledge on cloudflare workers while cloudflare blocks access.
the bridge solution (dev.to org) addresses this temporarily. long-term:
federated architecture means no single gatekeeper. multiple instances,
different providers, activitypub protocol
if cloudflare becomes restrictive,instances can move. thats the point of federation.
what would you prioritize for infrastructure that cant become a gatekeeper?
This actually hits a nerve in a good way.
What I like most is that you’re framing this as infrastructure, not nostalgia. Too many conversations stop at “Stack Overflow used to be better”, but that doesn’t help anyone. Treating this as a bridge phase makes sense. We dont get the perfect system first, we learn by publishing and seeing what actually sticks.
The focus on reasoning paths is key. Answers alone were never the value. It was seeing why something worked, why something failed, and how opinions shifted over time. That’s the part private AI chats completely erase right now.
I also think being explicit about verification and judgement is important. Junior devs aren’t just missing answers, they’re missing the muscle memory of doubt. Public discussion forced that, even when it was uncomfortable.
If this turns into a place where people document “here’s the decision we made and why we rejected the other options”, that alone would be hugely valuable. Especially in areas like architecture, security, and scaling where the wrong answer looks correct for months.
Glad to see people actually trying to build instead of just writing think pieces. Happy to follow along and contribute where I can.
this hits different. I've been noticing how much harder it is to find real problem-solving discussions now - everything's getting answered in private Claude/ChatGPT chats. the junior dev thing is real, like where do they even learn to debug when there's no public trail of "I tried this, it failed, here's why"? curious how you plan to incentivize people to post publicly when AI chat is so much faster and feels more private. also, the Stack Overflow decline is wild, I didn't realize it was 78%.
The point about developers solving problems in private AI chats really hits home. I've been seeing this firsthand — entire debugging sessions, architectural tradeoffs, security considerations just vanishing into chat logs nobody else will ever read. We're essentially privatizing the apprenticeship model that made the industry what it is.
Curious about the ActivityPub-based Q&A direction though. The hard part isn't the protocol, it's incentive design. Stack Overflow worked because gamification drove contributions. What's the retention hook here when people can just ask Claude instead of writing a public answer?
@richardpascoe done!
just removed you and sent a fresh invite
current status: Pending (waiting for you to accept)
check your email (spam folder too) for the DEV.to invite .should arrive in next few minutes .
let me know if the email doesn't arrive in ~10 min
Love the initiative and motivation. Keen to collaborate.
Love this initiative. Will be following this closely!
I do agree! Is there a way I can keep in contact (via email?) on more on the on boarding process? Let me know @richardpascoe @dannwaneri!
@francistrdev welcome! excited to have you join.
onboarding path:
@richardpascoe thanks for the intro Francis helping with testing is perfect timing since we need multiple instances to prove federation works
contact:
first goal: get 3 instances federating (you + Richard + Francis)
Sounds good! Going to take a look at it around evening (CDT time). Will let you know if anything! Thanks!