The Art of Engineering, or How I Learned to Stop Worrying and Love the Algorithm

Let me tell you A Story.

It’s 2007, and I’m sitting in my Brooklyn apartment, hunched over my laptop, scrolling through Tumblr. My cat, a moody tabby named Kafka, is perched on the back of my chair, judging me.

The screen glows with a chaotic mosaic of GIFs, poetry, blurry photographs, teenage bedroom art, obscure music recommendations and things that probably shouldn't have been interesting but somehow were. I’m not just scrolling. I’m discovering.

A seventeen-year-old in Ohio has posted a painting they made. Someone in Berlin has uploaded a playlist that feels like it was made specifically for me. A stranger has built a ridiculous little website about a band I've never heard of. Somebody has written 2,000 words about a movie nobody else seems to remember.

None of it has been optimized for me.

That's the point.

The early social web wasn't necessarily better. It was smaller, stranger, more chaotic and often painfully ugly. But it still felt like a collection of places. You wandered into them. You got lost. You followed links. You found somebody interesting and then followed the trail of breadcrumbs to ten other people you'd never heard of.

The internet felt less like a television and more like a city.

And somebody had to build that city.

The Internet Was a Creative Medium

We tend to talk about the early internet as though it was some innocent paradise before corporations discovered it. That's nostalgia talking.

The algorithms were already there. Google was ranking the web. Facebook's News Feed had already arrived. YouTube was already recommending videos. MySpace was already turning attention into a commodity.

The difference wasn't the existence of algorithms.

It was how much of the experience they controlled.

The early web still exposed some of its machinery. You could see the seams. You could make a website. You could change the background. You could inspect the source. You could build something ugly and strange and put it online.

The distance between having an idea and making a thing was sometimes ridiculously short.

That mattered.

HTML was not just a technical language. It was a creative material. CSS was not just a styling system. JavaScript wasn't just something developers complained about on Stack Overflow.

They were instruments.

A teenager with enough curiosity could turn a blank page into a world.

That was one of the genuinely radical things about the web: engineering and art weren't necessarily separate disciplines. The medium itself invited you to cross the boundary.

You didn't need permission to make the interface.

You could make the interface.

Then the Platforms Got Really Good

The modern internet solved a massive problem.

It became easier.

Much easier.

You no longer needed to know HTML to publish something. You didn't need to understand image compression to upload a photograph. You didn't need a server, a domain, a database or a terrifying conversation with somebody who knew what DNS was.

Platforms did something extraordinary: they turned the internet into an appliance.

And that was incredibly liberating.

Millions of people who would never have built a website could suddenly publish photographs, videos, music, jokes and ideas to millions of other people.

The trade-off was subtle.

We stopped building the places where culture happened.

We started renting them.

Your profile wasn't really your website. Your feed wasn't really your newspaper. Your audience wasn't really yours.

The platform owned the infrastructure, the interface, the recommendation system, the analytics and, increasingly, the relationship between you and everyone who came to see what you'd made.

The web became easier to use.

It also became harder to understand.

Enter the Algorithm

This is where the algorithm gets blamed for everything.

It's tempting to imagine some sinister machine sitting in a server rack, deciding which memes we're allowed to see.

Reality is both less dramatic and more interesting.

An algorithm is a piece of software designed to make decisions. Recommendation systems try to predict what you'll click, watch, listen to, buy or otherwise engage with.

They aren't evil.

They aren't benevolent either.

They're systems.

And systems have objectives.

If a platform's business depends heavily on attention, then keeping people engaged becomes an engineering problem. The system learns what holds attention and adjusts accordingly.

This creates a strange cultural feedback loop.

People make things for the platform.

The platform measures what works.

The platform distributes more of what works.

Creators notice.

Creators make more of it.

The system measures that.

Repeat.

Eventually, nobody needs to order culture around.

The optimization process does it for everyone.

That's the genuinely weird part.

No one has to decide that every thumbnail should have a shocked face. No committee has to decree that every video needs a hook in the first three seconds. Nobody has to announce that controversy performs well.

The system rewards behaviors that produce those outcomes.

And creators adapt.

When Everything Becomes Content

This is where something changed culturally.

A photograph stopped being just a photograph. It became a post.

A song became content.

A thought became content.

A person became a personal brand.

Even authenticity became content.

The strange irony is that the platforms didn't kill creativity. They industrialized it.

There are more people making things now than ever.

More musicians. More filmmakers. More illustrators. More writers. More bedroom producers. More weirdos with cameras and microphones and laptops.

That's fantastic.

But making something and making something that performs are increasingly treated as the same activity.

They're not.

Art asks:

What do I want to make?

The feed asks:

What will people click?

Those questions can occasionally produce the same answer.

They can also be completely opposed.

The Death of Sincerity Was Great for Engagement

This brings us to irony.

Irony isn't the villain either.

Irony is useful. It's funny. It's a weapon against bullshit. It allows people to play with meaning and say something without saying it directly.

The problem begins when irony becomes the only safe mode.

Online, sincerity carries risk.

Say something genuinely beautiful and somebody can make fun of you.

Admit that you care and somebody can screenshot it.

Make something earnest and somebody can turn it into a meme.

So we developed armor.

We say we love something “ironically.”

We post the joke before anyone else can make the joke about us.

We turn vulnerability into a bit.

We make fun of ourselves preemptively.

It's clever.

It's also exhausting.

And the algorithm doesn't particularly care whether you're sincere.

It cares whether you stop scrolling.

That creates an environment where emotional ambiguity can become a useful performance strategy. Everything can be simultaneously sincere and ironic, meaningful and meaningless, intimate and disposable.

Eventually you don't even know which one you are doing.

But Here's What We Get Wrong

The answer isn't to go back to 2007.

The old internet was not some utopian playground. It was inaccessible to plenty of people. It had terrible interfaces, broken links, harassment, misinformation and an astonishing amount of garbage.

And today's internet has accomplished things the old web could barely imagine.

A musician can release a record globally from a bedroom.

An artist can find an audience on another continent.

A teenager can learn Blender, Ableton, Python, animation, photography or electronics without paying tuition.

A person with an obscure obsession can find ten thousand other people who share it.

The problem isn't that the internet became too powerful.

It's that we increasingly experience its power through interfaces designed around a narrow set of incentives.

And that distinction matters.

Because it means the internet isn't finished.

Engineering Is the Medium

This is where engineering becomes interesting.

We usually imagine engineers as the people who build the machines that artists use.

But increasingly, the machine is the artwork.

The interface determines what can be expressed.

The database determines what can be remembered.

The protocol determines what can communicate.

The recommendation system determines what becomes visible.

The API determines what can connect.

The defaults determine what most people actually do.

That is not neutral infrastructure.

It is culture expressed as software.

A button is a design decision.

A feed is a design decision.

An infinite scroll is a design decision.

A notification is a design decision.

A ranking system is a design decision.

And every design decision changes behavior.

This is why engineering can be an artistic discipline.

Not because code is magically beautiful, although sometimes it is.

Because building systems means building the conditions under which other people think, communicate, create and behave.

That's an enormous creative responsibility.

Build Things Again

Maybe the answer isn't to abandon social media.

Maybe it's to stop expecting a handful of companies to build the entire cultural infrastructure of the internet for us.

Make a website.

Start a weird forum.

Build a tiny tool.

Publish an RSS feed.

Make a digital garden.

Release a piece of software that does one strange thing extremely well.

Create an interface that doesn't exist yet.

Build something that doesn't need to maximize engagement.

The interesting question isn't:

How do we escape the algorithm?

We probably can't.

The more interesting question is:

What can we build around it?

The first web generation taught us that a computer could be a publishing machine. The platform generation taught us that software could connect billions of people.

The next generation gets to decide what happens when we stop treating those two things as the same problem.

Maybe the internet doesn't need to become less technological.

Maybe it needs to become more human precisely by becoming more deliberately engineered.

The Playground Is Still There

Kafka is long gone now.

The internet I remember is mostly gone too.

But every once in a while, I stumble across something online that feels completely unnecessary.

A strange little website.

A ridiculous interactive experiment.

A song somebody made in their bedroom.

A digital artwork with twelve visitors.

A programmer who spent three months building something that probably should have taken a weekend.

And I get the same feeling I had back then.

Not engagement.

Not optimization.

Discovery.

That's the part worth saving.

The internet was never really the playground.

The internet was the ability to build playgrounds.

And perhaps that's the part we've forgotten.

The algorithm isn't the end of the story.

It's just another material.

So let's make something with it.