The Fertile Knowledge Season

Everyone has an AI take right now...

Investors, journalists, LinkedIn thought leaders, your company's CTO — they've all staked their claim. I've read plenty of them, and I don't have much to add to that pile. What I do have is a memory. And a name for this moment: The Fertile Knowledge Season. It won't last forever, which is exactly why it matters.

I'm a 40-something engineer who was a broke, curious teenager when the internet cracked open in the late nineties. I didn't have much going for me — wrong background, wrong zip code, no obvious path forward. But I had a modem and enough stubbornness to figure things out. And what I found on the other side of that dial-up screech changed everything. Not because of what the internet was, but because of what it temporarily erased.

It erased identity. It erased gatekeepers. For a brief, strange window, it didn't matter who you were or where you came from. I could strike up a conversation with a post-doc about something deeply specific, or end up in an IRC channel with someone from a band I loved, and nobody was performing for anyone. There was no algorithm deciding what the exchange was worth, no platform extracting value from the interaction. It was just people and ideas, colliding freely. That felt like oxygen.

If you were there, you know exactly what I mean. If you weren't, I'm telling you: it looked a lot like right now.

The Window Before the Playbook Gets Written

The early internet had a property that tends to get lost in nostalgia: it was genuinely, structurally egalitarian — not because the people involved were saints, but because the infrastructure hadn't yet been optimized for extraction. Nobody had figured out how to monetize your attention at scale. The ad-tech machinery didn't exist yet. The playbook hadn't been written.

That window didn't last. Advertising moved in, then the platforms, then the capital, and the commons got quietly enclosed. What started as a place where a kid from nowhere could learn anything and talk to anyone slowly became a system designed to keep you scrolling, clicking, and converting. The information was still there, technically. But it was buried under sponsored results and engagement-optimized content and the slow accumulation of friction that benefits the people who own the pipes.

I don't think advertising was the root cause so much as the first visible symptom. The deeper thing was that once someone figures out how to financialize a new medium, the medium starts bending toward finance. That's not cynicism — it's just pattern recognition.

I can feel the same arc beginning with AI.

"Alignment" and Other Polite Words

Right now, the gatekeepers are still catching up. The models don't know your credentials. They don't know your background, your institution, your tax bracket. You can ask a frontier model the same question a Stanford researcher would ask, and you'll get the same answer. That's a remarkable thing that we're already starting to take for granted.

But watch the language carefully. "AI alignment" is a real and important technical problem — keeping powerful systems from doing catastrophic things. It's also becoming something else: a framework for deciding whose questions get answered, whose use cases get prioritized, whose values get encoded as the default. The people making those decisions are not neutral. They are, almost by definition, the people who got there first with the most capital. That's not a conspiracy — it's just how enclosure works.

The early internet didn't have an "internet alignment" problem. It had an openness that felt naive in retrospect, but that naivety was load-bearing. It meant that a teenager from a humble family could learn to code by reading other people's source code. It meant that expertise was shareable before it was sellable. We are still, just barely, in that phase with AI. The question is what you do with it before it closes.

Life Finds a Way

Here's where a reasonable skeptic pushes back: the people who benefit from controlling this technology have learned from history. They watched the internet happen. They're not going to make the same mistakes. They're moving faster, lawyering up earlier, acquiring potential competitors before they become threats. The window will be shorter this time.

Maybe. I think about this as a pendulum rather than a ratchet. Nothing moves in only one direction forever. The enclosures happen, yes — and then the cracks appear. Open source keeps winning in places it shouldn't be able to win. Knowledge leaks. People build things in garages and dorm rooms that nobody saw coming. I've watched this happen enough times that I've stopped losing sleep over the long arc. History is full of things that weren't supposed to be possible.

What I'd push back on is the framing that it's adversarial. The people moving quickly to shape AI aren't villains — they're human beings doing what people do when a new frontier opens: trying to build something, secure something, make something last. That's understandable. The pendulum swings their way for a while, then it swings back. The ones who lose are the ones who weren't paying attention when it was their turn.

What I do think about is timing. There's a skill to knowing when to bask, when to hibernate, and when to get to work. Right now — this specific, weird, wide-open moment — is a time to work. The seeds were planted by researchers and tinkerers and open-source contributors who spent years doing things that didn't make obvious financial sense. That work is now sitting in front of all of us, accessible, waiting to be built on top of.

Don't Sleep On This

I didn't understand what I was living through in 1999. I just knew it felt electric. I knew I could learn anything if I was willing to be patient and a little obsessive about it. I knew that nobody was checking my credentials at the door, and that the quality of your questions mattered more than the name on your diploma. That's still true right now, and it won't be true forever.

The curious are going to outpace the certified. The people willing to ask naive questions of powerful tools are going to discover things that the credentialed experts will explain away until they can't anymore. The engineers who treat this moment as a playground rather than a threat will build the things that define the next decade — before the platforms calcify, before the paywalls go up, before someone figures out how to make you pay a subscription fee to access your own curiosity.

You don't need permission. You didn't then. You don't now. Curiosity is still the only credential that matters. It always was.