Becoming überdevs

Nietzsche has this inconvenient idea: that the will to power isn't a craving for a status/position, but the drive to become more fully itself - and that adversity, the unpleasant stuff, is precisely the whetstone. What does not kill me, and all that. The part we'd very much like to skip is how power is expressed. He called the one who overcomes himself the Übermensch - not a superhero, but a human who stops waiting for permission and starts legislating his own values.

So allow me a cheap and irresistible pun, because that's how I learned in 2010-era golden-age blogs: the developer who survives this particular decade won't be the one who out-LeetCodes the layoffs. It'll be the überdev - the beyond-developer - who climbs out of this codemonkey station and into the parts we sneer at: marketing, content, business.

GenZ drew a bad hand

We were chatting on Discord, I was showing Adan our progress so far. Adan is a typical Gen Z specimen - who got hit hard by the 2020s events (so far). First hit: the Pandemic during college: remote-first eroded entry-level paths, limited networking Second Hit: Artificial Intelligence: entry-level rungs vanishing. skill commoditized: craftsman apprenticeship is gone - the worse hit long term

The meme is true: Both boomers and zoomers ask millennials for help on troubleshooting their prints

A very hostile table to play...

Devs lost their market advantage

Coding used to be an arcane skill - and a high demand skill. If someone wants to build an application - he needs devs There was no way around it. Even if the build was a dev himself - the ask was too overwhelming for a single person. The exceptional success of solodevs running fundamental opensource projects is an amazing inspiration to all. However they are exceptional - the exception.

Now anyone with a Claude Code account can build. Especially typical web applications:

Hey, check what I built with Claude: http://127.0.0.1:8000

The scarcity of building ability is gone - now commoditized. (as long as the token costs are cheap as they are right now)

I don't think just devs were impacted by AI, all information/services professionals tend to be affected. But that's a whole new article - besides my point here.

A concrete study case

On another project, the main issues for user acquisition and retention was the lack of a native app.

It's been years the ~75% of humans surf the internet with mobiles - not desktop. Don't get me started on the +90% of traffic are bots, and why the internet is dead. Nor expose my views - like - mobile bad.

That's the hand we were dealt. Mobile is king - and we did not have an app. Devs were web/backend types - they weren't familiar with mobile stack. Resources were tight.

About a month ago the CEO himself - a clever guy with marketing background and curiosity, He asked: How hard is it? He installed Claude Code and started an app shell. He was stoked! In about 2 days - this app shell looks like a complete app. He was in ecstasy - as if he drank the mead of poetry.

The next week - the devs cobbled some api endpoints. Freaking done! A near-decade long gap closed - by a marketing guy with Claude Code In spite of least two previous unsuccessful attempts in the past, by previous technical people. He has done it!

Next month, Claudinho and team hammered out the rest.

Becoming an überdev

There was a wave of layoffs during the press industry crisis. Journalists got displaced by the internet content model. A common taunt against them was: "Learn how to code" This was certainly short-sighted advice - within a few years the tables turned. Developers are getting taunted "Learn how to weld" Perhaps we should start driving for Uber.

It's not like software engineering is solved. Fred Brooks' essential complexity is live and kicking.

The only way out is upwards - especially for GenZ This time to become something else, something at a higher level.

What the numbers actually say

As I was showing Adan the Google Analytics stats, expecting the usual story: a few hundred visits, everyone bouncing in eight seconds, the internet's collective shrug.

Instead the average engagement time per active user was 5m 03s.

Five freaking minutes! Multiples of current year attention span! The benchmark for "good" on the web is somewhere under a minute - most pages never see the far side of thirty seconds before the tab dies. And the calculators were even worse offenders:

The home page behaves like a normal web page - people glance and leave. The tools behave like an app. Seven minutes of someone actually sitting with a fuel-plan calculator, dragging sliders, doing the thing - Wow! I feel like a brainrot dealer!

My first instinct was the cynical one: it's me. Dev sessions leaking into prod analytics, a tab left open, hot-reload inflating the numbers. I went and checked. It wasn't - the seven minutes are real people from the interwebz.

The surprise

Here's the part that lands on the thesis. What holds them is the utility - the fuel plan that answers a question they actually had before a race, the gear calculator that tells them what speed they'll hold on a 20% gradient climb. That's software-driven-content.

It reminds me of the early PC days, when kids got a CD-ROM with an interactive multimedia encyclopedia.

To be frank - coding was the least in the value added chain - Claude Code wrote the first version in a few hours, we got it polished over a week. Anyone could do it. IMHO, what makes or breaks was domain knowledge - I love cycling, I'm a bit obsessed with bikes (building and upgrading). The engagement comes from knowing what cyclists agonize over, framing the tool around their problem, writing the thing so it's worth staying for. The überdev parts.

The numbers didn't say "good software". Bits-n-bytes was the cheap part; we must swallow our technical pride. Numbers says: "People value what you offer".