Confessions of an "Idea King"

What is happening inside my mind?

1:18 AM. The moment when most people are deep in sleep, but my mind decides it’s the perfect time to architect yet another ambitious project.

There’s a Chinese saying: “When three people walk together, one will always be my teacher.” But I’ve discovered a “corollary” to this wisdom—when I walk alone, my mind becomes schizophrenic and as if a committee of three lives inside myself, which is constantly generating ideas that range from brilliant to bizarre, often in the span of a single sleepless night.

In the past year alone, I’ve conceived, pursued, and in most cases, abandoned five distinct projects. Five. That’s the kind of output some small companies dream of achieving. If a startup could successfully implement and deploy five meaningful projects annually, they’d be scaling rapidly toward unicorn status. But here’s the uncomfortable truth: while companies celebrate execution, I’m stuck celebrating ideation.

The Gallery of Abandoned Dreams

Let me walk you through my year of inspiration:

1. The 1 AM “Resurrection”

Do you like the word “Resurrection?” I do, since that’s the title of Mahler’s 2nd symphony.

It started with nothing abnormal. I had deployed my deep learning course project on AWS—a tool accessible at www.nl2prot.org. The $15 monthly cost seemed reasonable until it didn’t, and I pulled the plug after two months to preserve my modest budget as a college student.

Then came the job posting. It emphasized MLOps, Cloud, Docker—all the things my dormant project represented. That night, sleep became impossible. At 1 AM, driven by an inexplicable urgency, I left my bed and spent 30 minutes bringing the application back online. The idea hadn’t just occurred to me; it possessed me. By 1:30 AM, my project was live again (and at least at the moment, live), and I could finally sleep.

Success rate: 1/1. The outlier.

2. The Solar Irradiance “Saga”

Do you like the word “Saga?” I do, since that reminds me of the old legends.

A blog post about hybrid time series forecasting methods caught my eye. The approach itself was straightforward — pair a basic forecasting model with an LSTM to learn residuals. Nothing groundbreaking. Yet somehow, my brain latched onto time series forecasting like a dog with a bone.

I rallied two university friends and pitched several ideas. We settled on solar irradiance forecasting with machine learning. My vision was grandiose: train a foundational model on large solar time series datasets to learn universal patterns, then fine-tune for specific locations. Transformers. Time series analysis. Foundation models. Every buzzword a resume could desire, all wrapped in one elegant package.

Reality, as it often does, had other plans. After implementing the Informer model, we couldn’t even reproduce the baseline results from existing literature. My friends suggested pivots—autoencoders, shifting the project focus—but the initial spark had faded. Two months after my “idea spur,” we quietly let it die.

Success rate: 1/2.

3. The Graph-to-GNN “Pivot”

Do you like the word “Pivot?” You’d better do, especially if you claim to be an expert in Excel.

My internship at the Leibniz Institute of Plant Biochemistry began with a clear mandate: design algorithms for analyzing large protein families. The research proposal suggested graph algorithms, so we dutifully pursued them. The results were biologically nonsensical.

At our dead end, inspiration struck again. Why not introduce Graph Neural Networks? It seemed brilliant until I realized I’d just volunteered to dive into one of the most complicated and contentious topics in biological machine learning.

After discussing with my supervisor, we pivoted to this new direction. I started with reinforcement learning (RL) on day one. By day two, I’d switched to evolutionary algorithms because my RL experiments weren’t working. During a team meeting, my supervisor remarked with barely concealed astonishment: “Yesterday you were working with RL, but today you’re playing with evolutionary algorithms…”

His tone wasn’t entirely critical, but the subtext was clear: I was moving too fast, pivoting too readily, chasing the next shiny approach before fully exhausting the previous one.

Success rate: 1/(2 or 3?), anyway the pattern is concerning.

4. The Citation Matcher That Never Was

Surprise? I won’t ask you that question for this one!

Stanford’s “LLM from scratch” course appeared in my feed. Within hours, I’d become fascinated—partially from genuine interest, partially from noticing how many job postings featured LLM experience.

Simultaneously, my PI at a US research lab was encouraging me to add more citations to a paper draft to preempt reviewer criticism. Then it hit me: there should be a tool that automatically matches papers from arXiv based on keywords, making citation discovery effortless.

I searched online. Nothing existed. The rush of discovering an unfilled niche was intoxicating. I called my friend immediately, breathless with excitement about starting this project together.

He seemed lukewarm. I was busy. The idea remains theoretical, filed away in the vast archive of things I’ll “definitely do later.”

Success rate: 1/(I don’t know).

5. The Data Visualization “Secret”

Well, I already start to regret not asking that question. Do you have a secret? I won’t tell you my answer to that, since one must keep the secret that one has a secret to keep.

Learning German through a specially designed Excel-based tool while simultaneously hunting for data science research internships created an unexpected intersection in my mind. I wanted something impressive on my resume showcasing modern data visualization tools like Tableau or Power BI.

The idea struck, and I’m writing this blog post mere hours after conception. Since I haven’t abandoned it yet, the details remain classified. I can only hint: it involves my experiences in Germany.

Success rate: TBD, but when there’s a will, there’s a way.

The Double-Edged Sword of Constant Inspiration

Some people view this quality as a researcher’s or developer’s greatest asset. Abundant ideas signal creativity, intellectual curiosity, and the ability to make unexpected connections. In academic and innovative circles, idea generation is celebrated, even fetishized. Brainstorming sessions exist specifically to cultivate this kind of thinking.

And there’s truth to this perspective. Many breakthrough innovations began as wild ideas that emerged at inconvenient moments. The ability to see possibilities everywhere in daily life, to connect disparate concepts, to envision what doesn’t yet exist—these are genuine strengths. In environments that value innovation, having a surplus of ideas is infinitely better than having none.

The creative process requires raw material, and I generate it in abundance. Each idea represents a possibility, a path not yet taken. Some of the world’s most successful entrepreneurs and researchers are known for their prolific ideation. They throw a hundred ideas at the wall, knowing that if even three stick, they’ll have built something remarkable. I remembered what Kaiming He, a prestigious researcher in deep learning, once acknowledged: 95% of the ideas fail, but with the remaining 5%, he became one of the most successful researchers in academia.

But here’s what the inspirational LinkedIn posts don’t mention: there’s a dark side to this gift.

The Graveyard of Good Intentions

Looking at my track record, only the first example succeeded. One out of five (maybe?). A 20% success rate would bankrupt most ventures. For every project that reaches completion, four others languish in various states of abandonment—some with code written, some with teams assembled, some existing only as excited late-night messages to friends who’ve learned to temper their enthusiasm. I sincerely apologize to everyone who was voluntarily or involuntarily involved in these projects…

The pattern is distressingly consistent: inspiration strikes at bedtime, urgency feels overwhelming, I commit energy and sometimes recruit others, initial obstacles appear, the spark fades, silence. My digital workspace is a graveyard of half-finished repositories, abandoned documentation, and TODO lists that will never be completed.

There’s a psychological term for this: “shiny object syndrome.” The constant pursuit of new ideas at the expense of completing existing projects. While I’m chasing the dopamine hit of a fresh concept, the previous idea—which seemed equally brilliant three weeks ago—collects dust.

The abandonment rate also carries hidden costs. When I involve others, like my two friends in the solar irradiance project, I’m spending their social capital along with my own. How many times will they respond enthusiastically before they start hedging their commitments? My supervisor’s astonished comment about my RL-to-evolutionary-algorithm pivot wasn’t just observation—it was a gentle warning about sustainability and focus.

Perhaps most troubling is the timing. These ideas consistently arrive when I’m trying to fall asleep. It’s now 1:18 AM as I write this, including the meta-idea to document this very phenomenon. There’s something almost compulsive about it, as if my brain refuses to rest until it’s generated and recorded the next big thing. Sleep deprivation doesn’t enhance creativity; it just makes me more susceptible to every passing notion feeling like genius.

Finding Balance: Channeling the Idea Storm

So where does this leave someone like me—and perhaps someone like you, if you recognize yourself in these patterns?

I’m also learning to distinguish between different types of ideas. Some are meant to be learning exercises—quick experiments to understand a concept or technology. Others are serious projects requiring sustained commitment. Confusing the two leads to either over-investing in sketches or under-investing in potential masterpieces.

Most importantly, I’m trying to embrace the uncomfortable reality that having fewer completed projects is better than having many abandoned ones. Quality over quantity isn’t just a platitude; it’s a survival strategy for minds that generate ideas faster than hands can build them.

Conclusion: The Idea Person’s Dilemma

Being an “idea person” is both gift and curse. We see possibilities everywhere, make connections others miss, and generate more potential projects than we could complete in a lifetime. This creativity is valuable—and I do mean that. The world needs people who can envision what doesn’t yet exist.

But creativity without execution is just daydreaming with extra steps.

I’m still figuring out this balance. My project success rate remains troublingly low, and I’ll probably conceive three new ideas before publishing this blog post. But awareness is the first step. Understanding that my midnight inspirations follow a pattern—that the urgency I feel at 1 AM might look different at 1 PM—helps me make better decisions about what to pursue.

To fellow idea people: your mind isn’t broken. The constant flood of possibilities isn’t a flaw. But it might need better flood management. Not every idea deserves the same investment. Not every spark needs to become a fire.

Sometimes, the most creative thing we can do is close the laptop, put down the phone, and let the idea wait until morning. If it still seems brilliant after a full night’s sleep, it might actually be worth pursuing.

And if not? Well, there’s always tomorrow night’s inspiration.

It’s now 2:31 AM. Thanks to AI, it didn’t take me long to convert a word document into a markdown file. I should probably take my own advice and go to sleep, although the idea of a second blog is starting to emerge in my mind. Maybe you will see that one in the next month.

Have you experienced the midnight idea epidemic? How do you balance creativity with execution? I’d love to hear your strategies—preferably during daylight hours.

By the way, Mahler’s 2nd symphony is epic. You should definitely give it a chance.