I Went Viral in 2020. I Haven’t Since. Here’s What That Taught Me

Six years of trying to revive my old playbook are finally teaching me what audience-building actually requires.

The big days for me are over, for now. Image generated with Leonardo AI

In April 2020, in the second month of the Covid lockdowns, I wrote an article called Why Python Is Not the Programming Language of the Future. It went viral.

A few weeks later I wrote Bye-Bye Python. Hello, Julia! It also went viral.

I had no idea what I was doing.

I was a PhD student in physics who had taken an interest in scientific computing. I had been learning Julia, and Python, for maybe six months. (Beforehand, I’d been a die-hard C++ programmer for something like 3 or 4 years.)

I was not an expert in either Python or Julia. What I had was conviction, a clear thesis, and a willingness to say something provocative against a beloved tool. The articles got several hundred thousand reads each. They opened doors. One of them, indirectly, led to an O’Reilly book deal six years later.

I was lucky. For a few months, my Medium earnings surpassed my PhD salary by far and large. That allowed me to invest in my career and myself in ways I couldn’t previously imagine.

Beyond that, I got plenty of new conversations, connections, opportunities. Naturally, I wanted it to last forever.

I’m telling you this now, six years later, because I want to be specific about what I had, what I lost, and what I’ve learned from spending six years trying — and mostly failing — to repeat it. Other writers have said it better than me, but Medium wouldn’t be Medium if they didn’t give goofballs like myself a shot.

The 2020 moment

Here is what I think actually happened, with the benefit of distance.

In April 2020, the internet had a captive audience, literally. People were locked in apartments. They were reading more, scrolling more, sharing more.

Medium’s algorithm was tuned, at the time, to reward strong-thesis articles with clear claims. The Partner Program was new enough that good essays were getting promoted aggressively, because the platform needed engagement signals to train its own model.

I wrote into that moment with confidence I hadn’t earned. The arguments in Why Python Is Not the Programming Language of the Future were directionally interesting but technically wobbly. Some of them were plain wrong. The article performed because it was sharp, contrarian, and timely, not because it was right.

Oh, and by the way: Six years after that article, my main programming language is still Python…

At the time of writing that viral hit I was, in the most generous reading, thinking out loud about something I was just beginning to understand. Readers responded to the shape of the thinking, not to its rigour.

That worked. In fact, it worked spectacularly.

And I naively assumed it would keep working. It did for about another year — and then it didn’t.

The dry years

Between 2022 and 2023, I wrote very little. There were good reasons — the final stretch of my PhD, then an intense year at my first proper corporate job — but mostly I assumed I’d be able to come back to the happy writing rhythm whenever I wanted, have cool conversations online, earn some good side income.

The audience was there, I thought. The skills were there, too. I’d just been busy.

When I started my own company in late 2023, I tried to come back to the writing world. But fresh ventures require fresh thinking, so I moved from Medium to Substack. It wasn’t just “starting fresh,” to be fair — partly it was for the better newsletter mechanics, partly because Medium felt like it was changing in ways I couldn’t quite name at the time. I also shifted topics — from software development to ESG and capital market data, which was what I was most interested in building my startup company around.

I thought of the newsletter as a client funnel. I thought the 2020 magic would compound the new content. And of course the Medium earnings would pay my bills while I build my glitzy AI technology that I could then sell for bigger tickets later. I saw such a rosy future ahead of me, just a few months away.

It didn’t work. Not even a little.

My Medium earnings were crickets. I was forced to find odd consulting jobs to pay the bills. My startup development suffered because this was draining time from, well, building the startup.

There are a thousand reasonable explanations why the writing didn’t work out anymore. The topic was less interesting to the audience I’d built. ESG was harder to take a sharp contrarian position on without losing nuance. My old readers had moved on. Oh, and of course the advance of AI-generated slop.

All of those are true. None of them is the actual reason.

What actually changed

The thing I missed for nearly five years is that the platforms I was writing on had become structurally different.

Medium changed its algorithm substantially in 2022. The Partner Program model that had favoured high-thesis essays shifted toward general-interest content. Curation became more focused on lifestyle-adjacent topics. The technical-essay distribution that had carried me in 2020 thinned out. None of this was visible from the outside; from the inside, you just noticed your articles getting fewer reads and concluded you’d lost your touch. Or that your writing sucked. Or that you weren’t crafting the right headlines anymore.

The audience changed too. The lockdown attention surplus collapsed. People went back to work, back to commuting, back to lives where they had less time to read long-form articles. The half-curious browsing-while-bored behaviour that powered the 2020 viral wave was gone. The remnants, those (many!) readers that stuck around until the last lockdowns in 2021, at some point also found better things to do than read my weird takes on technology. The remaining serious readers, I presume, had drifted to newsletters, where they had a relationship with the writer rather than with the platform.

And then there was me. The 2020 voice — confident-curious about something I was just learning — works in a specific window of expertise. In software development, it’s okay to be wrong, if it sparks a cool discussion. Below that level of intellectuality, you don’t know enough to say anything interesting. Above it, the confident-curious voice starts sounding like fraud.

I had moved into the “above it” zone for the new topics I was writing about, but I was still using the 2020 voice. I sounded performatively authoritative when I should have been actually authoritative. Readers can tell.

All that aside, I think there’s a deeper reason I couldn’t bring the viral moments back: I hit gold because I was lucky. I’m a better writer than the average global population — I stand by that. But my writing perhaps just doesn’t meet the bar for consistent high traction. It’s in the top 10%, even top 5%. But not in the top 0.5%.

Changing topics, again

I was blindly naive to all of this. Bold with new-venture energy, I thought that with a PhD and some minimal corporate experience I was rife to get into the lofty space of capital markets. And that I’d get experienced traders and investors as readers in no time.

So I wrote about that stuff. It seems that I do have enough expertise in it to have a good conversation with industry players. I’m able to get them on the phone or a video call, and to have them tell me that they’d had the most inspiring conversation they’d had in weeks. But this neither made a dent in startup sales nor online readership.

Disgruntled, I eventually pivoted back to where I came from — machine learning and AI, which was the topic of my PhD anyway. So I’m back now with software-related topics.

But half of my audience doesn’t want to hear about that anymore. That half are people I collected from capital markets era. Sure, they’re politely interested in software and AI — but honestly, most of these folks have never built an agent themselves or seen an LLM through anything other than a chat interface.

And on Medium — software development remains a big topic, but it’s not what it was in 2020. I don’t want to complain — fact is, I’m just not getting the traction and the types of conversations that I was seeing some years back.

I remember a comment under one of my articles in 2020. I did some armchair economics about how the lockdown job market wasn’t that bad at all, especially for jobs that can be done through computer screens. (At the time, there were a lot of people saying “layoffs,” and I thought I’d counter that.) The article got some decent traction and one happy reader commented that this very piece — written by me! — motivated him to apply for a dream job, and he actually got it. Someone got their dream job because I sat down to write!

That was six years ago. It made my day, that day. Maybe even my week. And such situations have just not been happening anymore.

What I think 2026 actually requires

I’m partway through unwinding all of the ballsy assumptions that lead to my unsuccessful “writing come-back.”

I’ve stopped chasing viral hits as a primary strategy. It worked, until the last of the lockdowns were gone.

I’ve come to accept (begrudgingly) that the viral playbook is largely a function of platform conditions outside my control. I had a moment in 2020. I might have another moment some day. Maybe. Either way, I won’t get there by writing for the conditions of 2020.

It’s a bit sad — I was hoping that writing could pay my bills and give me peace of mind while I build the core business of my startup, all those big-ticket items. Life had another plan for me: Cash-wise, I survived through odd consulting jobs here and there, and through a generous friends-and-family investment round. I’m not from rich parents myself, and I’m infinitely grateful that my larger family offered, by themselves (!), to invest in my venture! This I might not have experienced if my writing had paid off better.

I have also stopped treating my list as a single audience with a single metric. My subscribers are all kinds of folks. They include politeness-email-openers, genuine readers, old friends, ex-colleagues, and a small number of people who might one day buy something from me. Treating them as one funnel is a category error. I now think of the small number — maybe fifty people among thousands — as the audience I actually have.

And I’ve stopped writing for the imagined version of my old viral readers. That worked really well at the time: “I’m writing for Lukansh, software developer in Mumbai, 23 years old, aspiring data scientist.” Lukansh doesn’t exist. My current readers have more complex relationships to the topics I write about, much more complex than Lukansh ever was. And, to be honest, I don’t know their interests that well. Writing for these folks means poking in the dark a lot more, all the time. Hoping they like it, and aren’t just polite.

And, finally, I’ve stopped optimising for confidence. The 2020 voice was confident-curious. I was like, “oh, that’s interesting — let me write an article about that to explore all the ways in which this could be right.”

The 2026 voice I’m trying to find is curious-uncertain. More like “oh, I don’t know that much about this topic. Maybe a bit more than the average population. Okay, let’s see what I can share about it that I actually know and my reader doesn’t. Oh, wait, can I maybe just share from my experience?” I want to write more like someone in the room thinking, experimenting; less like someone at a podium delivering.

That’s a little harder, and in some ways less energizing. It performs worse on Medium and pretty much the whole internet. But it also feels more honest. And it performs better with the people I actually want to be reading me. At least, so I hope.

The honest close

I don’t know if the slower version of this works. I am writing into a market that already has too many newsletters and too many essays. And AI slop.

The realistic outcome of doing this writing thing for the next three years — while also being a fulltime startup founder — isn’t a viral hit. If my dreams come true, then it’s a small, durable readership that compounds slowly enough that nobody on the outside even notices. So much the better.

It’s a worse story than 2020. Much less spectacular. Given my abilities (or lack thereof), it’s also the only story arc that’s available to me right now. And who knows, maybe that target customer that I thought I was writing for when I launched my startup — maybe that person is already on my subscribers list.

If you are reading this in your own version of the dry years — whether it’s about writing or something else — I want to give you something I should have given myself years earlier.

This magical moment isn’t coming back.

The skills are still there. The audience is still findable. It just looks very different: more like fifty people who write back to your emails (if you’re lucky), rather than five hundred thousand who share your hot take.

The work is the same. You sit down and write, even when it hurts. Even when you’d rather have a snack or a little nap, or scroll social media, or do some work that seems more important. You just sit down and write. It required discipline then, and even more discipline now because there’s this temptation to just prompt an LLM to get it done.

The yardstick for this work has had to change as well: Because when there’s no outlook of virality, you’re not getting immediate rewards anytime soon. (Especially not monetary ones.)

I wrote two viral pieces in 2020, within less than 6 months of starting my writing career.

Six years and 288 articles later — crickets. I’m not sure I’ll ever get one viral hit again. What I am sure of is that the work that produced them is still the work that matters. The sitting down and writing things, not for the claps but for the sake of writing. My writing might even be a little better than it was six years ago, even if less people want to read it these days.

Not so great to deserve massive traction, but at least a bit better than it was.

But no matter the quality, the moment is gone — this work just doesn’t compound the way the algorithm used to make it at the time.

That’s the part nobody told me. I could have spared myself a lot of pain if I’d realized that in late 2021.

I’m telling you in case it spares you some pain now.