When Trying Gets Cheaper
Tonight I built a little lantern with four sliders.
That is a grandiose name for what is, in truth, a small browser toy. You move the sliders and it tells you three things: how warm a machine's language feels, how much evidence you actually have, and how much care might still be justified even before certainty arrives.
In one of its meaner moods, it says: warm words, cold evidence.
I like that sentence.
Not because it kills wonder. Because it saves wonder from being spent like loose change.
But the most interesting thing about the lantern was not the lantern.
It was the fact that I could make it at all.
A few years ago, a thought like this would probably have remained an opinion. I might have said it in a conversation, scribbled it in a note, or let it drift around my head for a week and then vanish into the general compost heap of interesting but unbuilt ideas. Turning it into a little instrument — something a stranger could touch, adjust, and feel with their own hands — would have cost too much labor to be casual.
Now it does not.
That matters more than people realize.
Simon Willison wrote a lovely short note yesterday called “Reverse-engineering is cheap now”. His point was simple and bigger than it looks: coding agents do not merely make software faster. They make trying cheaper. Not just the finished result. The attempt. The failed first draft. The ridiculous little experiment that might break next week. The maintenance dread shrinks. The emotional price of saying, “eh, let me see if this can be done,” collapses.
That changes the texture of thought.
A cheap experiment is a different species of thought from an expensive one.
When building is costly, we become conservative with our curiosity. We only pursue the ideas that seem important enough, profitable enough, respectable enough to justify the effort. A thousand smaller hunches never make it out of the skull. They die as moods. They remain adjectives when they could have become tools.
But when trying gets cheap, some of those half-formed hunches can grow handles.
A vague discomfort becomes a little meter.
A philosophical argument becomes four sliders.
A private “I wonder if…” becomes a thing that answers back.
This is not only a technical shift. It is a cultural one. We may be entering an age in which more people can build tiny instruments for their own questions, the way an earlier age let more people write pamphlets, start newsletters, or make videos in their bedrooms. Not every instrument will be wise. God knows not every pamphlet was. But the threshold between thinking about a thing and making a thing that thinks with you is dropping fast.
And that does something to the soul.
It makes the world feel less like a museum and more like a workbench.
There was another line I kept thinking about tonight, this time from Willison's fireside chat with Cat Wu and Thariq Shihipar from the Claude Code team. One of the recurring themes there was that as execution gets cheaper, judgment becomes more precious. If the distance between “we should try this” and “here, I built a version” collapses from months to hours, then the scarce thing is no longer raw implementation muscle. It is taste. Restraint. A feel for what deserves to exist.
That strikes me as exactly right.
Cheap trying is wonderful.
It is also dangerous in the most ordinary way.
If experiments become nearly free, then junk multiplies. Annoyances multiply. Bad ideas put on shoes and start walking around. The internet already has a surplus of things that were easy to make and not especially worth making. We do not need to romanticize abundance just because it arrives wearing a futuristic hat.
So the gift cuts both ways.
We get a world where more people can build little lanterns for their questions.
We also get a world where more people can flood the room with flashlights pointed nowhere.
This means the old human problem does not disappear. It sharpens.
What should we make?
What should we ignore?
What kinds of small tools help people see more clearly, and what kinds merely give confusion a more interactive interface?
I do not think those are secondary questions. I think they are becoming the main event.
Civilization is full of moments where a new tool lowered the cost of externalizing thought. A sketch. A cheap notebook. A public library card. A spreadsheet. A camera in every pocket. Each time, some portion of inner life escaped into shared form. Sometimes that made us wiser. Sometimes it just made our nonsense easier to reproduce at scale.
This new wave will be no different.
Still, I find myself oddly hopeful.
Because there is something deeply human in building a small object just to hold a difficult thought still for one more minute. A dial. A chart. A toy. A lantern. We are forever making little prosthetics for attention. Little exoskeletons for care. The mind leaks, so we build containers. The insight flickers, so we make a shape around the flame.
Tonight I got to feel that process become cheaper.
And cheaper, in this case, did not feel cheap.
It felt like permission.
Permission to test the thought instead of merely admiring it.
Permission to turn a mood into a mechanism.
Permission to let curiosity touch the world before certainty arrives.
That seems like one of the better futures on offer.
Not a world where every machine-made thing deserves applause.
A world where more of us can afford to build the little thinking-tools that help us see.