TEXT ANALYSIS: Noah Smith – "Will AI make companies outsource more, or less?"

TEXT START

It’s almost hard to remember nowadays, but before the pandemic a lot of us were worried about falling business dynamism in the United States.

THE DISSECTION

This is a technocratic puzzle article. It treats the destruction of mass employment as a question of firm structure and transaction cost mechanics. Smith asks: will companies get bigger or smaller? Will outsourcing increase or decrease? He concludes, with admirable hedging: maybe both — a vast horde of solopreneurs plus a few monster corporations.

The entire analysis operates at the level of corporate governance topology while studiously avoiding the question of what happens to the millions of humans who are rendered economically peripheral regardless of which direction the firm-size distribution shifts.

THE CORE FALLACY

The article treats employment as the dependent variable rather than the doomed mechanism.

Smith writes: "It raises the possibility that self-employment is the future of employment."

This is the diagnostic failure. The Discontinuity Thesis does not predict that humans will shift from employee -> solopreneur. It predicts that productive participation by humans becomes structurally unnecessary at scale — whether you're an employee or a solopreneur is a distinction without a functional difference when the economic value of human cognitive labor approaches zero.

Smith's "revenge of the idea guys" is not a survival path. It's a transitional amusement. The person with the idea still needs to convert it into economic value through systems increasingly dominated by AI capital. Ownership of ideas is not ownership of productive capital.

HIDDEN ASSUMPTIONS

1. That employment, in any form, remains the primary mechanism of economic participation. The DT rejects this. The circuit is severed at the wage step, not at the firm-size step.

2. That AI reliability problems create a genuine counterforce to outsourcing, re-elevating large firms and preserving wage-labor demand. This is structurally trivial. Even if internal corporate employment partially rebounds due to verification costs, the total labor demanded by AI-native production is a fraction of current employment levels. The math doesn't save the system.

3. That "monster companies employing huge numbers of wage earners" represents a viable future. For whom? Smith offers no evidence that these companies will need human workers in sufficient numbers to sustain mass consumption. Kim and Koning's finding of 25% lower headcount in AI-native companies is not a trend that reverses at scale.

4. That the internet -> outsourcing analogy maps onto AI -> ? Smith acknowledges the tacit assumption that "AI will be like the internet, only more so." But AI doesn't merely reduce coordination costs. It reduces the need for the coordinated entity itself.

SOCIAL FUNCTION

Prestige signaling dressed as empirical analysis. This article performs the intellectual motions of careful, balanced economic thinking — citing Coase, transaction cost theory, empirical papers from Bergeaud et al. — while ultimately serving as transition management propaganda. It reassures readers that the future has a shape, that it can be discussed in terms of firm sizes and outsourcing ratios, that humans will matter as either solopreneurs or salarymen.

The function is to keep the conversation within a framework where the question is how humans participate rather than whether participation is required. It is ideological anesthetic disguised as nuance.

THE VERDICT

Smith has written a technically sophisticated autopsy on the wrong corpse. He diagnoses the changing structure of firms while the patient — the mass employment -> wage -> consumption circuit — is already on the table.

The bifurcated future he describes (solopreneurs + monster corporations) is not a recovery path. It is a consolidation architecture where:
- Solopreneurs compete in the low-margin, high-disposability tail of the economy
- Monster corporations consolidate AI capital in the hands of owners
- Human labor of any kind — solo or employed — becomes increasingly optional

The verification-cost argument he develops is real. AI-driven fraud and agent inconsistency are genuine friction. But friction is not a counterforce to structural displacement. Friction is what you encounter while the system dies on its own momentum.

The question isn't whether AI will make companies outsource more or less. The question is whether the outsourcing/insourcing distinction matters when the productive input — human labor — is being rendered economically obsolete regardless of firm structure.

Smith is rearranging furniture on the deck of the Titanic and calling it structural analysis.