Indie Development Is Just a Math Problem

Three Stories

I recently came across a post on Xiaohongshu (China’s Instagram-meets-Pinterest) by an author called MI沐凡, about his friend running a cross-border e-commerce business in Shenzhen, selling USB cables.

This guy had a four-person “strike team” internally. A new product could go from concept to listing in as little as three hours. One person scraped trending elements from major platforms, another used AI to generate hundreds of product renders, a third wrote the copy, and the last one pushed everything to multiple stores simultaneously for testing.

What about the actual product? The physical thing?

No rush. They’d wait for the data to come in. Whichever listing showed signs of traction, they’d contact the supply chain—shipping within three days. Most listings went up and died quietly. Not even a ripple.

In one year, they put up over 3,000 listings. The ones that actually worked and made sustainable money? Fewer than ten.

But those ten sustained the entire company and bought him a new apartment.

Second story. Pieter Levels, known online as Levelsio, one of the world’s most famous indie developers. Nomad List, Remote OK, Photo AI—all built by him alone.

He’s publicly stated that he’s built over 70 projects, and most of them failed. He estimates each project has roughly a 2% chance of success.

2%. That means you’d need to build 50 projects before you’d statistically “deserve” one success.

Yet those few successful projects earned him millions of dollars per year.

Third story. Peter Steinberger, an Austrian software engineer and founder of PSPDFKit (successfully exited in 2021). After the exit, he didn’t retire. He went on a project-building spree.

Open his GitHub and you’ll see 50+ active projects—Peekaboo, VibeTunnel, Poltergeist, ElevenLabsKit… CLI tools, AI integrations, developer utilities, you name it. Eventually, OpenClaw (an AI agent project) broke through.

Three people, three completely different fields—e-commerce, indie development, AI applications. But look closely, and the underlying pattern is identical:

Ship a lot, hit a few, a few is enough.

It’s Really Just a Math Problem

Translate these stories from emotion to logic, and you get a simple formula:

Expected successes = Number of attempts × Success rate per attempt

Assuming each project has a 2% success rate (Levelsio’s empirical number):

1 project: 0.02 expected successes—virtually impossible
10 projects: 0.2 expected successes—still dicey
50 projects: 1 expected success—finally, hope
100 projects: 2 expected successes—now you have a safety margin

From another angle: how many projects do you need to launch before your probability of “at least one success” crosses 50%?

The answer is 35. (1 - 0.98^35 ≈ 0.505)

Want that probability above 90%? 114.

This isn’t motivational fluff. This is math.

The cable guy’s 3,000+ listings, Levelsio’s 70+ projects, Peter Steinberger’s 50+ repos—they may have never sat down and calculated this formula, but their behavior perfectly follows this mathematical logic.

Why Do We Only Swing Once?

If the math is this clear, why do most people give up after their first project fails?

Because we were raised on “elite thinking.”

Score first on the exam. Get it right the first time. Every shot should be a bullseye. We’re terrified of failure. We despise “wasted effort.”

This mindset works fine for exams, because exam rules are deterministic—you know what’s being tested, you know the correct answers, and there’s a clear causal relationship between effort and outcome.

But entrepreneurship isn’t an exam. The rules of entrepreneurship are: no correct answers, no deterministic causality, and most of what you do will fail.

Applying exam thinking to entrepreneurship is like navigating the ocean with a land map—the map shows roads, but you’re facing waves.

So most people end up like this:

Spend three months polishing a “perfect” product
Launch it. Nobody uses it.
Start doubting themselves: Am I not cut out for this? Am I just not good enough?
Quit.

They bet everything on a single attempt. Win and you’re a genius. Lose and you’re a failure.

But math tells you: losing is the norm. A 2% success rate means a 98% failure rate. Your first project failed not because you’re incompetent, but because that’s just how probability works.

How to Increase Your At-Bats

Once you see this as a math problem, the solution becomes clear: reduce the cost of each attempt, then keep swinging.

“Cost” here isn’t just money—it’s time and energy.

The good news? We’re living in an unprecedented era.

AI is cratering the cost of “building something.” An MVP that used to take three months can now be prototyped in three days with tools like Claude Code and Cursor. What used to require a designer, a frontend dev, and a backend dev can now be done by one person.

The cable guy’s strike team uses AI to generate product renders and lists a new product in three hours. Five years ago, this was unimaginable.

AI doesn’t help you build better—it helps you build more, and test faster.

This is AI’s greatest gift to indie developers—not helping you score 100 on a single project, but letting you go from building 1 project to building 10 in the same timeframe.

Attempts go from 1 to 10, expected successes go from 0.02 to 0.2. Add automation, template reuse, and rapid validation methodologies, and hitting 50 or even 100 projects isn’t fantasy.

Can You Improve the Success Rate?

Yes. But honestly, for people who haven’t had their first success yet (myself included), discussing how to improve the success rate feels hollow.

Because improving your success rate doesn’t come from reading books, taking courses, or studying methodologies. It comes from accumulated failure.

There’s a line from that Xiaohongshu post that captures it perfectly:

Every failed listing isn’t a failure—it’s telling me: this spot, the water’s too shallow; that spot, the fish don’t like this bait.

Listing A’s image had a high click-through rate—borrow that style next time. Listing B’s selling point got lots of inquiries—extract it and put it on every product page. Listing C’s negative review exposed a pain point—contact the supply chain for improvements.

Three thousand attempts aren’t three thousand repetitions—they’re a self-evolving trial-and-error system.

Each failure provides “calibration parameters,” making the next attempt slightly more precise. Going from 2% to 3%, 5%, 10% success rate isn’t achieved through epiphany—it’s earned one shot at a time.

So for those of us still at the starting line, the most honest advice is:

Chase volume first, then quality. Swing 50 times before discussing methodology.

You can’t learn to swim by reading about it.

Entrepreneurship Isn’t Gambling—It’s Rolling Dice

We’re too accustomed to describing entrepreneurship as “gambling”—betting it all, going all in.

But “gambling” implies you only get one chance. Win or lose.

That’s not how it works.

Entrepreneurship is more like rolling dice. Roll a 6 and you win. The probability of rolling a 6 on any single throw is 16.7%—not great. But roll 10 times, and the probability of hitting at least one 6 is 84%. Roll 20 times? 96%.

Roll once, you need luck. Roll a hundred times, you need math.

What you need to do isn’t pray for a 6 on this throw, but:

Find ways to roll faster (use AI to reduce development costs)
Find ways to roll more (automate workflows, validate quickly, abandon failing projects quickly)
Glance at the result after each roll, and adjust your technique (learn from failure, calibrate direction)

Then math will be on your side.

Write an article nobody reads. Build a product nobody buys. Send a tweet nobody replies to. It’s all fine.

Because every small, seemingly insignificant attempt is dropping a tiny pebble into the pool called “luck.”

Drop enough pebbles, and one day, the entire pool will be boiling—for you.