← Back to Library

The story of the mathematics of machine learning book

Tivadar Danka offers a rare, unvarnished autopsy of the modern creator economy, arguing that the most sustainable path for deep technical education isn't algorithmic virality, but direct community ownership. While many narratives celebrate the speed of viral growth, Danka's account reveals how a rigid, math-heavy focus on substance—rather than the fleeting hype cycles of crypto or generative AI—created a resilient business model that survived the collapse of its primary distribution channel. This is not a story about a lucky break; it is a case study in how intellectual rigor can outlast platform volatility.

The Architecture of Attention

Danka begins by rejecting the conventional wisdom that creators must chase the money or the algorithm. "The conventional advice is 'don't follow your dreams, do whatever makes you the most money.' But I'm not a conventional person. I walk my path, not others'." This framing is crucial because it sets the stage for a journey defined by friction rather than ease. He describes his early days on Twitter not as a marketing masterclass, but as a deliberate choice to treat the 280-character limit as a poetic constraint. "A 280-character limit per tweet is not just a restriction; it's a format that encouraged me to break down complex topics into short and clear chunks. It's almost like poetry."

The story of the mathematics of machine learning book

This approach stood in stark contrast to the prevailing content culture. Danka observes that "quality technical content was rare on Twitter. Most of the creators were riding the current hype wave to maximize engagement, hopping from topic to topic like skipping stones on water." He lists the shifting trends from frontend development to vector databases, noting, "I wanted none of that. Instead of shaping my content to my audience, I wanted to shape my audience to my content." This inversion of the standard creator playbook is the piece's intellectual core. By refusing to dilute his message for mass appeal, he inadvertently built a more loyal, high-value audience. Critics might argue that this strategy limits the total addressable market, but Danka's data suggests that depth creates a more defensible moat than breadth.

"I wanted to teach and lift up, not to appease."

The Fragility of Platform Dependency

The narrative takes a sharp turn when Danka addresses the sudden collapse of his primary revenue stream. The article details how the acquisition of Twitter by Elon Musk and the subsequent algorithmic changes "halted growth, buried threads, and crashed impressions." Danka notes that his revenue plummeted from a sustainable $1,000–$2,000 monthly to a precarious $300–$600 overnight. This section serves as a grim reminder of the risks inherent in building a business on rented land. He writes, "There's no algorithm to serve; it's just you and me. If you enjoy my content, you subscribe to receive all my posts. They won't be buried in a transient timeline feed; you can always find them in your inbox."

The pivot to Substack was not merely a change of venue but a fundamental restructuring of the creator-audience relationship. Danka highlights the stability of email as a distribution channel, contrasting it with the volatility of social feeds. However, this transition introduced a new set of challenges. The author admits, "I spread myself too thin. Even though The Palindrome was growing nicely... my progress on the book ground to a halt." This honesty about the trade-offs between content creation and community management adds significant credibility to his analysis. It underscores that no platform is a panacea; the burden of marketing and production often shifts rather than disappears.

From Digital Draft to Physical Legacy

The resolution of Danka's story lies in his decision to partner with a traditional publisher, Packt, to transform his digital drafts into a physical book. This move allowed him to offload the heavy lifting of formatting and production. "The amount of work hours they put into this book was incredible, turning my Jupyter Notebooks into a beautiful 700-page LaTeX project." The collaboration highlights a symbiotic relationship between independent creators and established institutions, challenging the notion that the two are mutually exclusive. Danka notes that the book's release in May 2025 was an "instant success," validating his years of labor.

Yet, the emotional weight of the piece extends beyond the mechanics of publishing. Danka dedicates the work to his mother, who passed away just as he began writing. "Everyone in my life tried to talk me down of my ambitious goals... Everyone, except my mother. She was the only one who knew that if there's a will, there's a way." This personal dimension transforms the article from a business case study into a meditation on persistence. It reminds the reader that the drive to create often stems from a deep, personal need to prove that difficult things are possible. The story also serves as a subtle counter-narrative to the tech industry's obsession with speed. While the industry raced toward agentic AI and vector databases, Danka spent four years mastering the fundamentals of linear algebra and calculus, proving that in an era of acceleration, patience is a competitive advantage.

Bottom Line

Danka's most compelling argument is that the creator economy's future belongs to those who prioritize depth over reach and ownership over virality. The piece's greatest vulnerability is its reliance on a specific type of high-intellect niche that may not be replicable for all creators, but its lesson on platform risk remains universally applicable. Readers should watch how this model of direct-to-consumer education scales as the market for specialized technical knowledge continues to fragment.

Deep Dives

Explore these related deep dives:

  • Acquisition of Twitter by Elon Musk

    The Twitter acquisition is described as a pivotal moment that crashed the author's revenue and forced them to migrate to Substack. Understanding the timeline, algorithm changes, and creator exodus provides essential context for this career-defining transition.

Sources

The story of the mathematics of machine learning book

by Tivadar Danka · The Palindrome · Read full article

Writing online about mathematics and machine learning started as a creative hobby for me, a way to recharge my batteries after cofounding a failed startup that burned my savings and two years of my life.

I have always had a passion for teaching. Even as an undergrad, I often found myself explaining the material to my coursemates. In my third year, I became a teaching assistant, teaching probability theory to students only one year below me. Later, as a researcher in a computational biology group, I used to give spontaneous lectures on (guess what) the mathematics of machine learning, sometimes even helping my grad student colleagues prepare for calculus and linear algebra exams.

The conventional advice is “don’t follow your dreams, do whatever makes you the most money.” But I’m not a conventional person. I walk my path, not others’. So, in 2021, I started a project with no promise of money, only a promise of personal fulfillment. A dream.

It turned out to be the most rewarding decision I have ever made.

Before you read on: we are running a fundraising campaign in December with a 30% forever discount for new paid subscribers. We are aiming for 1000 members, and we are halfway with 550 currently.

■■■■■▢▢▢▢▢ 55%

Every new paid subscription will contribute to the following milestones:

600 paid subscribers → Exclusive mini-course (Q2 2026)

The topic of the first mini-course will be “Mathematics of Machine Learning”.

700 paid subscribers → Another exclusive mini-course (Q4 2026)

A second mini-course with a different expert/angle. The idea is to slowly build an extensive “course library” for members.

800 paid subscribers → Dedicated Manim animator

We’ll bring on a Manimator so more articles come with clean, high-quality animations that make the concepts instantly click.

1,000 paid subscribers → Full-time writer

This is the big one. With a full-time writer on the team, we can ship more deep dives, more courses, and more structured learning tracks throughout 2026.

If you upgrade now, you’re not just supporting what exists today; you’re helping shape what we can build next year.

Back in early 2021, most of the machine learning community was active on Twitter, and I enjoyed spending time on the platform, feeling the pulse of the tech world. So, it was a good place to start writing online. (We are living a different life now. COVID, the war in Ukraine, and Trump: Season ...