The 'Good Enough' Threshold

In The You On AI Encyclopedia

The threshold is not fixed but domain-specific and moving. In software development, the good-enough threshold for routine tasks crossed in 2025; for architectural decisions, it remains distant. In legal work, contract drafting crossed; trial strategy has not. In medical diagnosis, pattern recognition crossed; bedside manner has not. Each domain has its own timeline, its own threshold, its own point where the economic logic of human expertise flips from worth paying for to not worth the premium. The aggregate trajectory is clear: the floor is rising across every domain of symbolic work simultaneously. What was specialist knowledge in 2023 is competent-journeyman in 2026. What is competent-journeyman in 2026 will be baseline-adequate in 2028.

Harari's framework locates this dynamic within the broader pattern of technological unemployment—but with a critical distinction. Previous automation displaced the routine; it could not touch the cognitive. The steam loom replaced the hand-loom weaver's motor skill, not the weaver's judgment about quality, color, texture. AI is the first technology that displaces judgment itself—or, more precisely, displaces the market value of judgment gradations below the good-enough threshold. Judgment remains; the market for it contracts. This is Segal's 'depth losing its market value' diagnosis—the observation that profound understanding becomes economically irrelevant when adequate understanding is cheap and abundant.

The psychological cost compounds the economic. When your expertise is devalued not because you became less skilled but because competence became cheap, the narrative of merit collapses. The meritocratic bargain—that years of disciplined skill-building would be rewarded with secure standing—is broken. Not by malice, not by injustice, but by efficiency. The market does not care about the sunk cost of your training. It cares about the cost of the output. And if a machine produces adequate output at fractional cost, the market's logic is inexorable.

Origin

The good-enough concept has roots in Herbert Simon's satisficing (1956)—the principle that bounded agents select the first adequate option rather than searching for optimal. Harari's application to the useless class is new, identifying the satisficing threshold as the point where human expertise becomes economically superfluous. The AI-era insight: the threshold is not fixed but rising—what counted as excellent in 2023 is merely adequate in 2026.

Key Ideas

Eighty percent quality, forty percent cost. For most commercial purposes, this is superior economics to one hundred percent quality at full cost. The remaining twenty percent stops justifying the premium.

Depth remains real, market value contracts. The senior developer's architectural intuition, the lawyer's instinct for weak arguments, the physician's diagnostic subtlety—these do not disappear. The situations where they matter shrink.

Threshold is domain-specific and moving. Each field crosses at different times; all are crossing. The aggregate trajectory is expertise devaluation across every domain of symbolic work.

Psychological cost exceeds economic. When competence becomes cheap, merit narratives collapse. The worker's sunk investment in training is not recognized by the market, which cares only about current output cost.

Feeds the useless-class mechanism. Not through sudden unemployment but through gradual undervaluation—still working, still producing, but producing output the market values less each year.