Coding is dead. Long live coding!

A few weeks ago I was walking through the streets of Avignon (south of France) with the best developer I've ever met.

He was a colleague on the Dataviz team at Datadog, my previous company, and he now works at one of the leading AI companies on the planet.

Of course we ended up talking about tech when our wives weren't listening :). And we landed on the exact same confession:

"Damn. I haven't written a single line of code in six months." 😳

So I asked him THE question. The one I now get almost every day:

Is it still worth learning how to code?

His answer came with zero hesitation: "Of course. It's never been more important."

That's my opinion too. So today here are my 2 cents on the topic, with a dataviz angle of course.

Code = power

That's the first important thing to understand. True yesterday, true today, true tomorrow.

Code is the only way to get what all of us want: to display exactly what we have in mind on a screen. No limitations, no compromises.

The problem is that it has always been hard to master. So we built workarounds.

No-code tools have been the big trend of the last 15 years. They are wrappers that hide the code from you. They make you 10x faster on a defined set of tasks, until you hit a wall.

Think Tableau or PowerBI: brilliant for data exploration, but it's 100% sure there is a specific need you won't be able to build at some point. That's why so many of my D3 students show up straight from those tools, frustrated by the ceiling they just hit (example).

The same thing happens inside the coding world. For instance, Shiny, Streamlit and friends encapsulate lower level code so you write less of it. Less code, less complexity, and also less possible customisation. Always the same trade.

To put it in a nutshell: code is and has always been the most powerful tool we have.

We have never written so much code

AI writing code does not make coding disappear. Quite the opposite. There has never been as much code written as there is today!

In dataviz, the pinnacle is JavaScript and its D3 library. It is the only tool that lets you build truly bespoke, interactive graphics. The only one.

And look at the trend!!!

I spend my days studying charts, and the shift is obvious from where I sit. The quality and the quantity of custom dataviz is skyrocketing. That represents billions of lines of code.

That's my second point: coding is becoming the norm for everyone, developers or not.

Coding is NOT writing code

"Ok Yan, but you just told us you haven't written code in six months."

True. 100% true. I don't type code anymore. I think, I architect, I plan, I prompt, I test, I read what comes back. The AI does the typing.

But typing was always only a fraction of the job. That part is gone. It's sad and it took me a while to accept, but it's true whether we like it or not.

Everything else is still very much here. Every single day I answer questions like:

• CSS tween or JS spring for this animation?
• Do I really need an external library here?
• Static site on GitHub, or do I need server side computation?
• Cookies or local storage to keep this information?
• Is this chart component something I'll reuse elsewhere? And if so, what's the smartest API for that function?

I could keep going for hours with this list. Not one of those questions gets easier by typing faster.

That's my point #3: we barely write code anymore, but there's still a lot to understand if you want to work well with AI.

So, should you learn?

Yes.

Code is the most powerful thing you can put in your hands, it has never mattered more than today, and it comes with a whole body of knowledge that sits on top of actually typing the characters.

For a dataviz practitioner, learning that knowledge is the single best investment of your time. "It opens a new world" like several of my students framed it.

⚠️ But not the way we used to learn it.

Learn the concepts, not the syntax.

It doesn't matter if you can't write a forEach loop from memory. You don't need to memorize the options of D3's linearScale.

What you do need is a deep understanding of the pillars of programming. The foundation. And that's exactly what I teach in my d3 loves react project.

Otherwise you'll end up not understanding the joke below 😀

The bottom line

I'm an AI skeptic, in the sense that I don't think it's great progress for humanity.

But for us dataviz lovers, it's a huge step forward, and I can already see it in the quality of the work being published.

In an era where misinformation is everywhere, the mix of code, AI and dataviz might be our best shot at fighting back. So let's make good charts the norm!!

PS: here's the viral post I made on the topic.

Let's talk soon,
Yan