Creating software at human scale

We spent most of the last decade of our digital lives on large platforms built by cloud companies.

“All of the apps are built,” a friend and I recently observed. There are no more opportunities to be the next “Uber of XYZ.” If there is not already an app for it, it is already a part of some other platform.

And yet, we aren’t satisfied. Platforms track us constantly. Sure, you can use an alternative browser and lock down your privacy settings. But social media finds ways of profiling us anyway. It’s all built with algorithms we cannot see, touch, or adjust.

Algorithms are all about scale. Not just scaling hardware, but scaling our attention as well. Whether it’s Meta, Amazon, or Airbnb, it’s built to scale to millions or even billions of people. All while maximizing the time, attention, and money we spend.

Why? These companies are built for “web scale.” Years ago, “web scale” was a line from a silly video about MongoDB. But it is also the most succinct way of describing the mess of our current software landscape.

The Supreme Court heard arguments this week about social media moderation laws. Underneath it all, there is a debate about scaling. Meta has billions of users across its platforms. You can connect to nearly anyone in the world with a phone. It is massive.

And for this reason, it does not want to be responsible for content it can’t moderate. It doesn’t scale. Even with the help of automated systems, it can barely keep up. But if it doesn’t do some moderation, its platforms suffer. They become a swamp of spam, disinformation, and extremism.

Even with moderation, there are so many users on Meta’s platforms you often can’t tell what is real and what is fake. Our very minds have a difficult time keeping up with it all. People are often misunderstood. And if you aren’t playing to the algorithm, you aren’t getting heard at all.

Where are we going from here? Will there be another platform around the corner? Will it promise fixes for everything wrong with the old ones, only to get scaled up as well? Will we give up?

We are ready for something different. Coincidentally, we are also witnessing a transformation of the software industry through AI.

It is now possible to try multiple solutions much more quickly than before. Problems previously too expensive to consider solving are now within reach. But most importantly, serving niche needs cost effectively is now possible.

Successful software engineers in the next decade won’t be the mythical “10x developer.” We won't need as many who scale from hundreds of thousands to millions and billions. Instead, excelling at breaking down problems and organizing information will be key. They won’t be “frontend engineers”, “backend engineers”, or even “full stack” engineers. They will be “complete product” engineers, able to deliver from concept to completion.

This will all mean more software created for highly specific needs and communities. Software that does not algorithmically abuse our attention.

Software created for human scale.