Creativity is a matter of attempt count, not talent. Today we unpack this through trying 3D coding on a free AI.
Let me say something about creativity. What separates creative people from the rest? Talent? Inspiration? Neither. The gap is number of attempts. And number of attempts is inversely proportional to experiment cost.
Today's example is having a free AI write 3D code. Not practical. Not even good results. But cost was zero, so I could try ten times. Three years from now, when this free model is gone, the principle still works.
Simple math. Each try costs $10 — how many tries? Maybe one or two. Each try costs $1 — ten. Each try costs $0 — a hundred.
Why does this matter? Because unexpected discovery hides inside one of those hundred. Someone with one or two tries never touches it. Someone with a hundred tries touches it. The real secret of creativity is here. Not talent — attempt count.
One day I asked a free AI (DeepSeek) to code a 3D game. A foolish attempt. I knew paid Opus would do it better. But it was free, so I tried.
Try 1. Error-ridden. Fail. Try 2. Runs a bit. But weird. Try 3. Different approach. Fail. ... Try 10. Surprisingly pretty result. Different from my original intent, but better.
Had I run this on paid Opus, I'd have stopped at try one with "eh, good enough." Free meant zero cost of abandonment, and with no abandonment cost, exploration kept going. Try ten was the real discovery.
When cost approaches zero, the horizon of exploration changes.
To feel this, picture a painter. One painter owns one expensive brush. Another owns ten cheap ones. Who experiments more with stroke variety?
The cheap-brush painter. No fear of ruining. They spill black ink, soak a brush in water, dry it and scratch with the bristles. Inside that mess, their own stroke appears. The expensive-brush painter never does such experiments. Only one brush.
Ever wondered how a creator's "own style" forms? From the experiment count of cheap brushes. Not from expensive ones.
Once the principle is in you, daily strategy changes. "How can I lower the experiment cost?" starts running in your head.
Lower cost → lower fear → more tries → surprising discoveries. This virtuous loop is creativity.
Don't use this principle in reverse. "Real work needs an expensive tool" — the opposite is true. Expensive tools are for verified output. Exploration comes from cheap tools.
Try one hundred times on free AI, and when a promising direction shows up, finish on paid. This order works. Reverse it and you just repeat finishing without exploring.
This sounds abstract. Numbers. I ran the experiment myself.
For two weeks, I built the same project (a short interactive web page) two ways.
Method A — paid Opus only. Perceived cost per try: $0.50. Tries in two weeks: 6. First approach was decent, so I kept refining. Result: clean but plain.
Method B — free models for exploration, Opus for finishing. Per-try cost: $0. Two weeks: 47 exploratory tries + 3 finishing tries on Opus. Most of the 47 were garbage. Then around try 31, a strange animation appeared. "What is this?" I took it to Opus and finished. Result: unique, memorable.
6 attempts vs 50. That gap separates "plain" from "memorable." Not skill — attempts.
The principle works even harder for beginners. Beginners don't know "which direction is right," so the exploration itself is the learning.
Experts guess right 50% of the time on first try. Beginners, 5%. What do beginners need? 10× the attempts of an expert. Expert's 5 tries = beginner's 50 tries to reach the same outcome. More tries.
Lower cost accelerates beginner learning. Learning Claude Code, I ran over 300 attempts on free models. 290 failed. The patterns from the 10 wins became my skill. The 10 wins were impossible without the 290 fails. Failures are the raw material of skill.
Three immediate moves:
Creativity isn't talent. It's a function of attempt count. And attempt count is a function of experiment cost. Lower cost, more tries. More tries, more surprise discoveries.
Three words to carry — Cost. Try. Discover. Lower the cost, raise the tries, wait for the discovery.
This isn't only about AI experiments. Writing, painting, cooking, business, exercise — same. One expensive attempt loses to a hundred cheap ones on creative output. Whatever tool comes next, this principle repeats. Technology changes. The principle of attempts doesn't.