As the proud owner of seven Pomodoro timer apps, I can confirm Iām excellent at starting things and terrible at finishing them. The real grind begins after the excitement wears off and youāre left alone with your own half-finished ideas staring back at you. If youāve ever shipped code, printed a carved block, or tried to get a toddler to nap, you know the last 10 percent is where chaos hides.
I named this newsletter The Last 10 Percent because thatās the part of work I canāt stop thinking about. Itās the part that separates a clever prototype from a product someone trusts. The part that turns an idea into a habit. The part that keeps artists and engineers up at night.
Iāve been a software engineer long enough to know this pattern by heart. Well, my 15 year career has always lived in this 10 percent space. The happy path works perfectly. Then you ship it, and reality shows up with a list of ways users will break your assumptions. The real work, the hard work, the work that separates āit works on my machineā from āit works everywhere,ā happens in that final stretch.
Iāve also been a human long enough to know that life works the same way. I carve linocuts in my spare time, and itās a similar story. The easy parts come early, the last bits like the repairing, refining, showing up again when youād rather be done, is what really builds character. Iāll be writing about that in all forms i.e. the messy middle of making things, the slow lessons of parenthood, and the quiet work of becoming a little better at being human.
In tech, thanks to shiny new AI tools, you can make something impressive quickly, but productionizing it is a different species of effort. The logs get weird. The metrics contradict each other. Itās the model that works perfectly in a notebook but falls apart once it meets real-world latency. The test suite that passes locally but fails in CI because the environment isnāt quite the same. The fine-tuned LLM that looks great in evaluation until you realize it hallucinates 1 in 20 times, and no oneās sure why. Itās feature flags that donāt roll back cleanly, flaky integration tests that hold up a deploy, prompt chains that break when one dependency updates.
I think these are the same problem wearing different clothes.
Sometimes the last 10% isnāt even about code. Itās about the slow, patient work of making something accessible, maintainable, and kind to whoever inherits it next.
Thatās the kind of invisible grind Iāll be exploring here.
Most content about software development focuses on architecture, new frameworks, shipping fast. Most content about creativity focuses on inspiration, getting started, finding your voice. But the space between āalmost doneā and āactually doneā? Thatās where I keep finding myself stuck, learning, and having the occasional āaha momentsā worth sharing.
So thatās what this newsletter is about. A messy mix of essays, experiments, and whateverās rattling around in my brain that week. Each issue will poke at this finishing stretch wherever it hides: shipping a product, showing your work even when itās not pretty, wrestling with a deployment that refuses to cooperate, re-alignment of linocut blocks and prints, tinkering with a video edit, or finally finishing a print thatās been haunting my desk.
And while I love AI (I have more AI side quests than I can count), this space will stay free of AI assistance and ads. A small corner of the internet made for humans, by a human. Me.
I think the last 10% is where the most interesting work happens. Itās just that nobody wants to admit theyāre still there. Letās change that.
P.S. If youāre wondering whether this newsletter itself will suffer from last 10% problems (typos, inconsistent publishing, occasional rambling that needed more editing), the answer is yes. Weāre learning in public here. Welcome to the mess.