Where Do We Stand on the Technology Wave?
We are living through the largest information breakthrough in history. Larger than the one Gutenberg triggered when he bolted a wine press to a set of movable type. And what were the consequences? Religious reform, social restructuring, mass job obsolescence. Scribes in Paris attacked and destroyed presses out of fear. And yet, by 1500, printing presses operated in 282 cities across Europe, Venice alone running 150. Dittmar (2011) found that cities adopting the press grew 30 to 60 percentage points faster than those that didn't. Plenty of jobs died. Far more were created.
Saturn Devouring His Son, Francisco de Goya (1820)
My sister sent me a paper recently. Descriptive, and it matches my anecdotal view of the market. Its thesis is simple: entry-level tech jobs are evaporating. The industry that built the machine is the first to be fed into it. Saturn devours his children. Evolution at its best. When a paradigm shift happens, the community of the affected domain splits in two: those who are reluctant to accept the new paradigm, and those who embrace it. We all know what happens to the former. Their lenses become obsolete to analyze the phenomena, because the new ones are so much better.
The Analogy
Hegel wrote that peoples and governments have never learned anything from history. He was describing our default mode. Learning from history is hard because you can't replicate experiments the way you can in the hard sciences. But it is possible to look at patterns. The scale of what's happening is unique, but the structure is very similar.
Niall Ferguson has been making this argument for years. In The Square and the Tower he frames the printing press as a network revolution, same as the internet and AI now. Ada Palmer takes it further: the computer, the internet, social media, and LLMs are not separate revolutions but deeper penetrations of one. The press wasn't done reshaping the world after 1455. It kept generating successive waves for 150 years. Money moves fast. Institutions move slow. How long will ours take?
The printing press, at a smaller scale, teaches something precise: when a new information technology arrives, the scribes eventually become obsolete. But the editors become a new guild. The bottleneck moves, therefore the sources of value.
The Scribe Fallacy
What I think is happening is what I'll call the Scribe Fallacy. You have ten engineers. A new tool arrives that makes each engineer 10x more productive. The manager's response? Fire nine. Keep the best one. Promote him to Head Scribe.
Think about how absurd this is. It is the year 1455. You are watching Gutenberg's press spit out pages. And you conclude that what you need is one really fast scribe. This is a comprehension failure at the most fundamental level. Today it seems obvious that you want multiple editors working together. In 1455, it was not obvious at all.
The managers think they understand the new paradigm. They do not. They are looking at a printing press through scribe-colored glasses. They see a tool that makes scribes faster. What they should see is a tool that makes scribes obsolete and editors essential.
Printing press, woodcut by Jost Amman (1568)
The Gun and the Soldier
There is a Dunning-Kruger epidemic around these tools. People use ChatGPT for a week and crown themselves AI engineers. They've discovered fire. They are, in their own estimation, unmatchable.
This is probably how the first person to fire a gun felt. Point, click, the other guy falls down. Who needs years of sword training? Who needs military discipline? The weapon does the work.
But you know how this story ends, because you live in a world shaped by centuries of evidence. A trained soldier with a gun is standard deviations better than me with a gun. The weapon didn't eliminate the need for skill. It moved the skill. It raised the ceiling and lowered the floor simultaneously, because the sample grew significantly.
Full Metal Jacket, Stanley Kubrick (1987)
The Contrarian Position
I'm going to say something that sounds trivial, and somehow it is the key argument for my contrarian position. These tools are massive value creation machines. Any role can be improved by significant standard deviations if you give that person agentic coding capability. Not just engineers. A marketing manager who can spin up A/B test pipelines without waiting for dev. An ops lead who builds internal dashboards in an afternoon instead of filing a Jira ticket. An HR team that automates candidate screening, onboarding flows, and comp benchmarking in days, not quarters. A finance analyst who builds real time forecasting models instead of copy-pasting into spreadsheets. Every department, same pattern: the bottleneck was always "waiting for someone technical to build it." That bottleneck is dissolving. But this is not a weekend trick. It is a full time discipline. Mastery requires dedication. If you want to be at the top, it requires obsession.
So fundamentals are still relevant. What is changing is the discipline. Painting used to be scientific inquiry. When Galileo pointed his telescope at the moon in 1609, he drew what he saw in sepia wash. Thomas Harriot had observed the same moon through a similar telescope months earlier, but his sketches were crude. Galileo could read the craters because he had trained in chiaroscuro at Florence's Accademia delle Arti del Disegno. His artistic literacy was a scientific instrument. Then photography arrived and painting became a wildly inefficient way to model reality. Did something get lost? Of course. A photograph doesn't capture what a painter's eye selects and emphasizes, and the understanding of an object is not the same after drawing it by hand as after simply taking a picture. But the opportunity cost doesn't justify the method. And painting didn't die. It became art, a luxury, a niche. The default mode of production moved on. The same thing is happening now. A financial analyst building models by hand when an AI does it in seconds. A lawyer drafting boilerplate when a language model produces first drafts in minutes. A marketing team waiting three weeks for a developer to build a landing page. The craft isn't wrong. The method is becoming obsolete.
Some information is lost by distancing from the handcrafted detail, but the abstraction is worth the time. Each layer builds on the previous one. We stand on our forefathers' shoulders not because we learned their craft, but because we trusted their work enough to build higher. We trust our phones, our medicine, our legal system without understanding what is going on under the hood. Trust in abstraction is the default mode of civilization. That is what agentic coding enables. You don't need to write every function from scratch. You need to know what to build, how the pieces come together, and why you are adding value.
Becoming an end to end software engineer used to be something very senior. Today it is going to become standard. Not only getting the backend and frontend right, but also having the right design taste and, I'd argue, the business perspective and the right copy voice. Solopreneurs will rise. They will have issues. New bottlenecks, just as editors did.
If I were in a position of power, my bet would be simple. Find the high agency techies who are willing to put in the work. The ones who are able to automate and improve other people's jobs, instead of leaving the people who need their job automated by the best engineer. Stick to people who are building with it, daily, pushing into the boundaries of what these systems can do. The people who treat this not as a parlor trick but understand it as a new paradigm, agentic coding.
Two groups disagree with my position on still betting on software fundamentals. One with words, the other with action. That is what makes it contrarian. The first tells you to stop learning to code. Jensen Huang at NVIDIA, Amjad Masad at Replit, Sam Altman at OpenAI. Every one of them has an army of top software engineers. Model competition is fierce. Imitating the state of the art gets cheaper by the month. They do not need a market with more software developers. The second fires nine, promotes the survivor to Head Scribe, and wonders why quality collapsed. The first plays fool, the second pretends to know. The tango is the same. Fundamentals. Hard work. The paradigm of our time.
Galileo Galilei, original drawings of the Moon (1609), Biblioteca Nazionale Centrale, Florence