April 2026 · GoCodeLab · Dev Mindset

Why I Started 30 Side Projects and Finished 3

In my Notion there's a page called App ideas to build someday. Inside that page there are 47 cards. On my laptop there's a folder called side-projects/, and inside it there are more than 30 folders. Of those, the ones I shipped to launch number three. The first commit on the other 30 is almost always stamped around 2 a.m., and the last commit dies on average 17 days later. The folder names all sound plausible. focus-timer, meal-log, retro-cli, kkulpot-v2, idea-board. Thirty headstones. The epitaph on every one is the same. The first week was actually fun.

Am I lazy? You can't really call someone who codes until 3 a.m. on the first week of a new project lazy. Then is it ADHD? I suspected that for a while too. Every item on the self-screening checklist looked like a description of me. To put the conclusion up front: I spent way too long stuck on that distinction. To be more honest, the distinction itself doesn't matter much. What matters is which environment you're inside, and only that.

Dopamine Seeking vs. ADHD — Does the Distinction Even Matter?

Search and you'll find self-screeners and I check every box. Good at starting, bad at finishing. Hooked easily on something new. Can't stand a boring task. I used those to suspect myself of ADHD for quite a while.

Then I read a 2009 paper by Volkow and colleagues. Evaluating Dopamine Reward Pathway in ADHD, a PET imaging study peering into the dopamine system of adults with ADHD. The conclusion was the opposite of what I'd assumed. The availability of D2/D3 receptors in the reward circuit was lower in the ADHD group than in controls. It isn't that they have too much dopamine making them scattered — it's that the dopamine signal is weak, so they keep hunting for stronger new stimuli. I had to stop and re-read that part.

So ADHD and the dopamine-seeking pattern start at opposite points but land on the same behavior. Both chase new stimuli. ADHD chases bigger ones because the signal is weak. Ordinary dopamine seeking gets learned because the environment hands out stimuli too easily. Both output as create another new folder.

What I came around to much later is this. The two aren't a clean either/or — they're a spectrum. One end is clinical ADHD. The other end is a person with almost no dopamine seeking at all. Most indies, including me, sit somewhere in between. Where you sit isn't decided by a single self-screening checklist. It's decided by whether your daily functioning shows clear impairment. Building 30 new folders is not impairment. Losing your job, severing every relationship, not being able to run your day — that's impairment. Those 30 headstones are just the output of your environment.

(Of course, if you genuinely feel you meet the clinical criteria, that's a real psychiatric evaluation. This article isn't here to diagnose anyone. The point is: don't pathologize yourself based on the 30 started / 3 finished pattern alone.)

How a New Folder Triggers a Dopamine Spike

That novelty triggers dopamine is essentially settled in neuroscience. Animals that responded strongly to new environments, new food, new mates were the ones that survived. The dopamine system evolved to reward exploration. More precisely, when you explore, it makes you predict that a reward is about to come. That prediction is what moves the animal. Dopamine isn't the chemistry of arrival — it's the chemistry of pursuit.

How does this apply to indies? The dopamine that fires when you create a new folder is not coming from a finished app. It's coming from the prediction that the app will be finished. Five minutes after you hit mkdir new-thing, your head sees a future where the app is launched. You see a tweet that says "this is genuinely good." You see 50 users hitting your first paid checkout. Those scenes look real. You haven't written a single line of code. But the dopamine fires off that prediction. And dopamine response drops on familiar stimuli. Which is why, in week three, your codebase no longer feels new. And right then, in the shower, a new idea shows up. The circuit closes.

There's another layer. In 2009, Kashdan and Silvia wrote a chapter in the Handbook of Positive Psychology called Curiosity and Interest: The Benefits of Thriving on Novelty and Challenge. It frames curiosity not as a byproduct of happiness but as an engine of it. People who actively pursue new stimuli scored, on average, higher on subjective well-being. Higher on life satisfaction. Higher on meaning-seeking. Curiosity isn't a defect. The same impulse that made you start 30 projects is part of the circuitry that keeps you alive.

Place these two facts side by side and you can see yourself clearly. You're a dopamine-pursuit animal, and the pursuit is what keeps you going. That pursuit hits the environment of solo dev and keeps leaking out as another new folder. The circuit isn't broken. The output shape is just expensive for you.

Honestly, the Most Absurd Pivot I Ever Made

One time, I was a week away from launching the app I'd been building. Nine of 12 beta testers said it was good. All I had to do was hit the launch button.

That night, on my X timeline, I saw a toy project someone had built with Cloudflare Workers + R2. A stack I hadn't used. I read their GitHub for 30 minutes. Then I created a new folder. worker-storage-toy. The app that was a week from launch got pushed back a week that night. That week became two months. Because I tried to learn Cloudflare Workers and integrate it into my actual app. It didn't integrate. It didn't fit my app's stack. Two months later I dropped worker-storage-toy, came back to the original app, and shipped it.

It's embarrassing, but it took me a long time to admit that I didn't do it because Cloudflare looked good. I did it because the dopamine on my own app had cooled, right before launch. Right before launch is when dopamine cools the most. You're almost done, the novelty is gone, and what's left is debugging and store metadata. That's when your brain starts demanding a new stimulus. The X timeline conveniently delivered one. I didn't betray my app — I got dragged by my dopamine system. Two months disappeared. Zero dollars to the user, two months from me.

But the 30 Aren't a Total Loss

Read this far, and the 30 sound like pure waste. There's one thing I shouldn't leave out.

In reinforcement learning there's a concept called the exploration vs. exploitation trade-off. Do you explore new options, or exploit the option you already know is good? You need both. Pure exploration never harvests a result. Pure exploitation never finds a better one. Optimal is some mix.

My 30 headstones are an exploration cost. In one of those folders, I tried SwiftData for the first time. In another, I learned about PostHog's free tier. In another, I met the component library I eventually plugged into my main project. Those folders didn't ship themselves, but they were quietly building the parts that ended up inside the three I did finish.

That's why Kashdan & Silvia call curiosity a psychological resource. Curiosity looks like a one-shot cost, but over a lifetime it stocks an inventory you can pull from. Those 30 headstones are my warehouse. The parts in that warehouse make it more likely the 31st project finishes faster. I didn't make 30 projects — those 30 made me.

You can't use this as an absolution though. To turn 30 into inventory, you have to write down at least one line each — where I stopped, what I learned. I often didn't. Which is why half of my 30 are just two months that vanished, and only the other half became inventory. That half is what built the three I shipped. The next 30, I want all of them to become inventory, even if I only write a couple of lines per dead folder. That's one improvement I can actually make.

Practical Strategy: The 48-Hour Idea Queue

One prescription. Willpower can't beat dopamine. You have to change the environment.

The rule is simple. When a new idea shows up, write it only in your notes for 48 hours. Don't hit mkdir. Don't search for a domain. Don't create a GitHub repo. Just the name, a one-line description, and why you thought it was good. Close it. Open it again 48 hours later.

Why 48 hours and not 24. At 24 hours, the dopamine of the new stimulus only cools slightly. At 48 hours, it cools enough that you're reading the note in a sober state of mind. From experience, that gap matters a lot. Some notes are still attractive at the 24-hour mark and turn ordinary at 48. Choosing not to start the ordinary ones saved me a huge chunk of time. About 95% of my notes turn ordinary at the 48-hour mark. The 5% that survive are my real candidates.

Lay one more small rule on top of this. Inject novelty into the current project on purpose. Rewrite a single component with a new library. Drop an unfamiliar tool into one small endpoint. Do it inside a corner of the current project. Don't make a new folder. A new folder resets sunk cost to zero. That's the dopamine vending machine. A new library inside the existing project gives you the dopamine and keeps the sunk cost. Your brain accepts that as I explored, and your project doesn't die.

One last line. I still want to make new folders. Even as I write this, two new side projects are floating around in my head. They aren't going away. As long as I live as an indie, that circuit will stay on. The difference now is that I wrote those two onto my App ideas to build someday page. I'll look at them again in 48 hours. If they turn boring, that's one fewer headstone. If they don't, that's one more real candidate. The 30 headstones weren't proof I was broken — they were the output of my environment. Change the environment, and the output changes. That's how the 31st becomes the finished fourth.

This article reflects information as of April 2026. The cited studies don't change with time, but my interpretation of them might shift with better data.