There was a time when progress meant working in the dark.

Not metaphorically, literally. Before the light bulb, before electrification, before the infrastructure we now take for granted, innovation looked like long stretches of uncertainty punctuated by small, hard-won breakthroughs. Edison did not invent the light bulb in a straight line, and he did not have a clear roadmap or a set of best practices to follow. What he had was persistence, and a willingness to try far more variations than most people would tolerate. More than 1,000 attempts, most of them failures.^[1]

That was the cost of invention: time, materials, and a tolerance for getting it wrong.

AI feels like that moment again, not because we are reinventing the light bulb, but because we have dramatically collapsed the cost of trying. What used to take weeks can now take hours. What used to require teams can now be done solo. What used to be blocked by access to knowledge, to compute, to tooling, is now sitting behind a browser tab and a bit of curiosity. The barrier between idea and experiment has thinned to almost nothing.

I felt that shift myself while prototyping a small tool to auto-clip coupons for a grocery chain.^[2] It never promoted to a full product, and that was fine. The point was to see if it could exist, and how quickly I could get there. A year ago, I probably would not have started. Now, the only real cost was a bit of time and curiosity.

When the cost of experimentation drops like that, the entire landscape changes. It starts to feel less like a structured field and more like a frontier. There is no single "right way" to build with AI right now. No stable patterns, no time-tested playbook, just a growing collection of people wiring things together, testing assumptions, and sharing what worked (and often, what did not).

It is messy. It is inconsistent. It is occasionally ridiculous.

The Wild West had its share of those, too. Not every bold idea paid off; just ask the investors behind pneumatic mail systems that attempted to move packages through underground tubes across entire cities. The vision was ambitious. The execution did not quite hold up. But that was the pattern: big swings, frequent misses, and the occasional breakthrough that made the rest worth it.

If you zoom out, this is what a creative explosion actually looks like. Not polished or orderly, but full of parallel experimentation, most of it going nowhere, some of it quietly changing everything.

Edison's advantage was not just intelligence. It was throughput. He tried more things. He ran more experiments. He failed faster and more often, which, counterintuitively, put him closer to success. That kind of iteration used to be gated by resources and time. Now, it is mostly gated by initiative.

AI has not guaranteed better outcomes. It has done something more subtle, and more powerful. It has made it easier to attempt better outcomes. That means we should expect a flood of low-quality, half-working, quickly abandoned ideas alongside a smaller but meaningful set of breakthroughs that would not have existed otherwise. More noise and more signal. More failure and more invention.

That tradeoff is not a flaw. It is the mechanism.

This leaves us in a moment that rewards participation more than perfection. You do not need to have the right idea, and you do not need to wait until the tools stabilize or the patterns settle. By the time they do, the frontier will have already moved. The people who benefit most from shifts like this are not the ones who observed them best; they are the ones who engaged early, even if what they built was messy or incomplete.

The gap between thinking about something and trying it has never been smaller. That is the opportunity. It is also the challenge, because it removes the most common excuse: waiting until you know more.

Somewhere right now, someone is building something unpolished, slightly unhinged, and unexpectedly effective, maybe even in Steve Yegge's Gas Town, not because they had a perfect plan, but because they started, tried something, and kept going.

Most attempts will not go anywhere. Some probably should not. But every now and then, one does, and that is enough to justify the rest.

That is the game now.

Try more things.