Over the last few weeks I’ve spent a surprising amount of time rebuilding something that almost nobody in 2025 truly “needs”: the UNIX.com man pages app. I moved the stack from Ubuntu 18 to 24, cleaned up Apache and redirect rules, tightened the Rails app, wired Redis properly, and added a simple, fast global man-page search. Redis is now running at over a 96% hit rate. Technically, it’s the cleanest and fastest this system has ever been.
And yet, if we’re honest, the world doesn’t need another man-pages site. It doesn’t need another classic forum either. Most people now ask an LLM instead of reading man ls or opening a thread on a UNIX forum. So why is unix.com still here, and why did I bother upgrading it instead of shutting it down?
This is about more than a stack upgrade. It’s about what it means to maintain an artifact from a fading era, long after its peak usefulness to society has passed.
The Technical State: A Polished Relic
On the technical side, the man-pages app is now in good shape:
- The OS is upgraded from Ubuntu 18 to 24.
- Rails is tuned for read-mostly access.
- Redis caching drives a >96% hit rate.
- The old slow index that depended on heavy DataTables JS is gone.
- The main listing for man pages per OS is now a “fast index” – simple SQL, aggressive caching, and clean URLs:
- /man-pages/:os(/:page)
- Search is intentionally minimal: an exact match on the man page name (query), returning OS, section, hits, and a link. No fuzzy search, no full-text noise, no query-string gymnastics. Just “show me all the awk man pages across all OSes.”
I deliberately avoided chasing “perfect Rails.” No obsession with the asset pipeline, no overuse of helpers, no trying to make everything a textbook example of Rails orthodoxy. In many places I just dropped inline CSS and explicit href="/man-pages/linux/3" links. It’s faster to build, easier to understand, and less brittle than abstracting everything in the name of framework purity.
So from a systems perspective: the app is lean, predictable, and low-maintenance. It behaves like a well-tuned archive, not a fragile science project.
Frameworks, Dogma, and the Buddhist Raft
This is where the classic Buddhist parable of the raft shows up in software engineering.
You build a raft to cross a river. Once you’re on the other side, carrying that raft on your back everywhere you go is pointless. It was useful, then it became baggage.
Modern frameworks and “best practices” often behave exactly like that. They help you move from chaos to order. After that point, endlessly hauling around all the theoretical perfection – full asset pipelines, ideal REST shapes, ultra-DRY abstractions – just becomes attachment. You end up serving the framework instead of the actual purpose of the application.
In the man-pages app, I consciously dropped a lot of this “raft weight”:
- I didn’t optimize for the perfect Rails pipeline.
- I didn’t force everything through helpers when a hard-coded path was simpler and more robust.
- I chose minimal diffs over clever abstractions. If a small, direct change solved the problem, I took it.
It’s a very literal example of letting go of dogma and focusing on what matters: the app is fast, stable, and simple to reason about. That’s enough.
A Quiet Community and a Long Tail of Gratitude
This is not a shutdown announcement. The site is still running, and will continue to run. But given how much the world has changed, it’s a good moment to say something that doesn’t get said often enough: thank you.
unix.com has been around for decades. Over that time:
- Many thousands of users asked questions, shared answers, and discussed about the right way to do things.
- Moderators quietly cleaned up spam, kept discussions civil (most of the time), and enforced standards long before “content moderation” was trendy jargon.
- Past admins and technical contributors helped migrate servers, upgrade software, fix security issues, and keep the lights on when it wasn’t simple or fun.
If you ever posted a question, answered one, reported spam, suggested a fix, or helped in the background: you had a hand in making unix.com what it was at its peak and what it still is now – a high-signal UNIX/Linux resource that doesn’t insult the intelligence of its users.
These contributions don’t vanish just because traffic is lower today. The content is still there. The decisions are still encoded in the archives. People still land on those old threads and man pages and find something useful. And yes, machines crawl it too and quietly absorb what was written.
So to everyone who ever invested time, energy, or patience in this place: thank you. You helped build something that lasted far longer than any of us could have predicted when it started.
LLMs, Forums, and Who This Is Really For Now
In 2003–2010, the answer to “who is this for?” was obvious. People came to unix.com to ask real questions and learn. Man pages were primary documentation. Forums were live communities. You could feel the traffic and the energy.
In 2025, the landscape is different:
- Most casual and intermediate questions are answered by LLMs, StackOverflow remnants, vendor docs, or some Discord.
- Many younger engineers never touch a real man page. They ask a model instead.
- Traditional forums like
unix.comare largely obsolete as “front-line” tools.
So who’s using it now?
- A small number of humans, in specific cases:
- People on older systems.
- People who still prefer canonical references.
- People who stumble onto a man page or thread from a search result.
- And a lot of machines:
- Search engine crawlers.
- Indexing systems.
- Training corpora for current and future LLMs.
In practice, unix.com and the man-pages app are now part of the substrate the modern AI ecosystem quietly feeds on. The world doesn’t see that; it just sees “ask the machine a question and get an answer.” But under the hood, archives like this are part of the fossil record that models learn from.
So yes, in 2025, I’m maintaining something that is arguably more useful to machines than to humans. But it is still useful, and it still carries the fingerprints of everyone who contributed along the way.
Attachment, Legacy, and Not Letting Go (Yet)
If I’m honest, unix.com remains online for one main reason: I’m not ready to let go of it yet.
The site is tightly bound to my own history: early TCP/IP deployments, Air Combat Command networks, years of UNIX and Linux work, and decades of community Q&A. Turning it off isn’t just flipping a switch on a website; it’s closing a chapter of my life.
From a Dharma perspective, I can see the attachment very clearly. The world doesn’t need this site. The forums are a shadow of their peak. The man pages are one of many reference sets. But I still keep it running because it’s part of my story, and because it still quietly serves a few humans and a lot of machines.
At some point in the future, it may make sense to shut it down as a symbolic gesture – not because the hardware fails or I’m forced to, but as a conscious act: an acknowledgment that the era of classic forums and man-page-driven learning has effectively passed, and that LLMs and other systems have taken over much of that role.
When that day comes, it will also be a chance to say a final, explicit thanks to everyone who helped build this place. For now, this is just an interim reflection and a nod to the people behind the content.
Why Keep It Running When the Future Is AI?
We just completed a major upgrade: Ubuntu 18 to 24, app refactoring, routing and caching improvements. From a purely rational standpoint, this would have been a perfect moment to shut everything down cleanly if that’s where my heart was. Instead, I made it faster and simpler to keep online.
The truth is straightforward:
We all understand the future is AI, not forums, blogs and man pages.
People will keep typing questions into models, not into old bulletin boards. That’s fine. But those models still learn, directly or indirectly, from the kind of material that lives here: technical discussions, man pages, code snippets, real operational experience. The front-end has changed; the value of solid, well-curated reference material hasn’t vanished, it’s just less visible.
So for now, unix.com and the man-pages app remain online. Not because the world “needs” them in the way it once did, but because:
- The system is clean, fast, and low-maintenance.
- It still quietly serves a few humans and more than a few machines.
- It represents a long chapter of our own lives and work that I’m not ready to erase with a single DNS change.
At some point in the future (2035, in 10 years from now perhaps?), there will be a time to archive it properly, make one last clear statement, and turn off the dynamic parts as a symbolic gesture that this era has truly passed. That day is not today.
Today, it’s enough to say: the future is AI, but this small corner of the old world still runs, still answers, and still stands as a trace of the culture that helped build the systems everyone now takes for granted.