Were Software Engineers blue-collar workers all along?

TEXT START: There’s a common idea among economists that you can broadly split employees into two categories.

The Dissection

The article is trying to demote software engineering from a privileged professional identity to a skilled production trade. That reframing is partly accurate: developers build and repair things, and coding involves craft, practice, and operational competence.

But the article’s real function is psychological containment. It converts an ownership and labor-demand crisis into a tooling problem: learn the new machine, remain useful, and everything will be fine. The analogy comforts software engineers by placing them inside a familiar history of technological adaptation.

The Core Fallacy

It confuses being skilled with being economically indispensable.

Blue-collar workers did not generally “ride” technological change safely. Mechanization repeatedly reduced labor demand, destroyed bargaining power, and concentrated ownership. The surviving workers were often those needed to supervise, maintain, or operate a larger automated system—not the original workforce preserved at its former scale.

AI is also not merely a better loom or chainsaw. It targets the cognitive production process itself: drafting, debugging, documentation, testing, estimation, and increasingly system design. The relevant question is not whether engineers can use AI. They can. The question is whether one engineer using AI can replace ten, fifty, or five hundred engineers. If so, tool adoption accelerates labor substitution rather than securing employment.

The article mistakes continued human capability for continued human market value. Those are different things.

Hidden Assumptions

• Productivity gains will produce enough new software work to absorb displaced developers.
• Firms will distribute AI-created surplus through wages rather than retain it as profit or transfer it to asset owners.
• Human judgment will remain scarce at roughly current staffing levels.
• Managing automated systems will require large crews rather than a thin supervisory layer.
• Historical technological transitions are reliable guides to general-purpose cognitive automation.
• Enjoying coding as a craft has economic relevance after coding ceases to be well paid.
• “Use the tool” is a survival strategy even when every competitor has access to the same tool.

None of these assumptions is secured by the article. Several are precisely what the Discontinuity Thesis rejects.

Social Function

Partial truth serving as ideological anesthetic and transition management.

The article correctly identifies that software engineering has always contained a production-trade element and that developers are not immune to technological displacement. Its failure is more consequential: it treats adaptation as preservation. It tells a threatened labor aristocracy to accept lower status and greater output while leaving ownership, bargaining power, and labor allocation unexamined.

The “manager of robots” image conceals the arithmetic. A warehouse requiring one manager and one hundred robots is not a victory for the one hundred warehouse workers replaced. It is a smaller labor market with a higher return to whoever owns the robots.

The Verdict

Software engineers may have been blue-collar workers in the sociological sense, but that classification misses the terminal mechanism. The decisive distinction is not white collar versus blue collar. It is owner versus displaced operator.

AI does not need to make programmers useless to destroy their privileged position. It only needs to make competent software production abundant, cheap, and scalable. Learning the loom may preserve a niche for some engineers—especially those who own systems, control capital, verify consequential outputs, or maintain scarce infrastructure. It does not preserve mass employment, former salaries, or the postwar wage-consumption circuit.

The honeymoon is not ending because software engineers have finally discovered they are tradespeople. It is ending because the machines are beginning to perform the trade.