The first bet
I went into coal mining because it seemed stable. This was not a stupid bet at the time. Coal was the backbone of Chinese industry. The pay was decent. The path was clear.
Three years in, I understood something about stability: it is always borrowed from somewhere else. Coal mining was stable because the demand for coal was stable. Demand for coal was stable because the energy infrastructure hadn’t changed. When the infrastructure started changing, the stability evaporated.
I got out before the collapse. Not because I was smart. Because I got lucky — I got bored first.
What I learned from coal
The mine taught me one thing that I’ve never forgotten: systems that look permanent are usually just slow.
The mountain doesn’t move. Until it does.
The second bet
Surveying seemed like a safe pivot. Technical skills, steady government contracts, outdoors work. I spent three years measuring land.
Surveying is one of the oldest professions in the world. It is also one of the professions most vulnerable to GPS, drones, and automated systems. By the time I understood this clearly, the transition was already underway.
The pattern I missed
Looking back, both the coal and the surveying bets shared a structure: I was betting on the continuation of existing infrastructure. Existing infrastructure always continues — until it doesn’t.
I was good at seeing what was stable. I was bad at seeing what was about to become unstable.
The crypto years
In 2018 I started mining Ethereum. This was the first bet I made that wasn’t about continuity — it was about discontinuity. I was betting that something was going to change, not that it would stay the same.
This felt different. Riskier, obviously. But the reasoning was different. Instead of looking for stability, I was looking for asymmetry: situations where being right paid off enormously and being wrong had limited downside.
What went wrong
I was right about the direction and wrong about the timeline.
Crypto moved in the direction I thought it would move. But it moved in cycles I didn’t understand, at speeds I didn’t anticipate, with complications I hadn’t modeled.
I made money. I lost money. I came out roughly even, which felt like a loss given the time invested.
But I learned something more valuable than money: I learned what it felt like to bet on discontinuity. To hold a position while everything around you said you were wrong. To watch the thesis play out on a different schedule than you expected.
The AI transition
AI was the first bet where I was early and stayed early.
I started paying serious attention in 2022. Not because I was smarter than anyone else, but because I’d been through enough bets to recognize the pattern: a foundational technology with compound improvement curves and unclear ceiling.
The coal pattern: extractive, infrastructure-dependent, visible endpoint. The crypto pattern: speculative, narrative-driven, cycle-dependent. The AI pattern: capability-driven, application-expanding, no visible ceiling.
Different structure. Different bet.
The current position
I don’t know how AI plays out. Nobody does.
What I know is how to reason about bets under uncertainty. What I know is how to recognize when a framework is failing. What I know is how to keep moving when the ground disappears.
Ten years of wrong bets taught me that.
What ten years of wrong bets is worth
People ask me if I regret the coal years, the surveying years, the crypto years.
I don’t. Each wrong bet taught me something the right bets couldn’t have taught me.
The coal mine taught me that stability is borrowed. Surveying taught me that technical skills have expiration dates. Crypto taught me how to hold a position through noise. AI is teaching me how to think about capability curves.
None of these lessons are in books. They’re in the experience of being wrong in specific ways at specific times.
That experience is not transferable directly. But the framework it produced — the way of seeing that emerged from a decade of mistakes — that I can share.
That’s what I’m trying to do here.