Taste Is All That’s Left – How AI‑Generated Code Shifts the Engineer’s Craft

TL;DR

AI has made generating plausible code cheap, so the only remaining scarce skill for engineers is taste—the personal, hard‑to‑define judgment about what is truly good enough and worth shipping.

The Wall Has Been Rented Out

The author observes that the traditional barrier between idea and working program—hours of typing, debugging, and learning APIs—has collapsed. Modern LLMs can produce a plausible implementation from a natural‑language description in seconds. This shift does not erase the value of the experience gained while climbing that wall; it merely relocates it.

"You can describe a thing now and receive a plausible version of it much faster than you could have typed the first function by hand." – Taste Is All That’s Left

Why "Good Enough" Is a Problem

The essay calls “good enough” a solvent that dissolves the incentive to improve. When production was expensive, effort acted as an invisible filter, ensuring that only ideas that survived the cost of creation reached users. With friction removed, the filter disappears, and the market no longer rewards the extra effort of discerning quality.

"Effort was a filter, and like all filters it was invisible until it was removed." – Taste Is All That’s Left

Defining “Taste”

The author distinguishes taste from mere preference. Drawing on Robert Pirsig’s notion of Quality, taste is described as a wordless, compressed verdict that arises before we can articulate a reason:

"Taste is the compressed, wordless verdict you reach faster than you can justify. It is the ‘no, again’ you say to yourself with total conviction and no available argument." – Taste Is All That’s Left

Taste is therefore the only part of software craftsmanship that cannot be automated.

How Friction Built Taste

According to the essay, the difficulty of building software served as a curriculum:

"The friction was not an obstacle to developing taste. The friction was the curriculum. Every wall I cursed while climbing it was, without my noticing, teaching me which walls were worth climbing." – Taste Is All That’s Left

Repeated failures forced engineers to internalize patterns of good and bad design, creating an accretive sense of judgment.

The Economic Consequence

When taste is present, engineers ship at the same speed as those without it, because the extra time spent rejecting plausible but sub‑optimal solutions is invisible to market metrics. This creates a hidden cost:

"Taste does not show up in the diff. It is unmeasurable, uncreditable, and invisible on a dashboard." – Taste Is All That’s Left

Consequently, engineers who maintain high standards incur personal overhead while the market rewards raw output velocity.

Community Reactions

@boron1006 notes that LLM‑generated code often lacks signal, forcing developers to read hundreds of words to understand a module.
@mdwelsh echoes the sentiment, saying his own hard‑won taste makes him skeptical of AI‑built demos.
@jdzikowski frames taste as a curriculum learned under scarcity, now shifted to adoption.
@madrox argues that AI shortens the half‑life of taste, turning it into a commodity rather than a competitive advantage.
@agentultra emphasizes that taste follows experience and suffering, not a single “no‑again” decision.
@burnto points out that while anyone can make anything, only those with taste can judge what is worth keeping.
@sbloz raises the practical problem of curation: with an infinite flood of plausible outputs, how do users discover the scarce signal?

These comments reinforce the central claim: AI removes production friction but does not replace the need for human judgment.

The New Craft: Choosing Over Building

When manufacturing became cheap, thinkers like Ruskin asked whether everything should be made. The modern parallel is that choosing which AI‑generated artifacts to keep is now the primary engineering activity.

"The scarce act is no longer making. It is choosing. Deciding what, out of the endless generated plausible, deserves to exist and be kept." – Taste Is All That’s Left

Practical Takeaways for Engineers

Invest in personal judgment – Treat taste as a skill to cultivate, not a nostalgic relic.
Use AI as a tool, not a replacement – Prompt LLMs, then apply your own “no, again” filter before shipping.
Document decisions – Since taste is invisible to dashboards, write explicit design rationales to preserve institutional knowledge.
Focus on curation – Build processes (code reviews, style guides, internal libraries) that surface high‑taste work amid the flood of AI‑generated code.
Accept the hidden cost – Recognize that maintaining quality may slow you down, but it remains a differentiator that cannot be outsourced.

Conclusion

AI has democratized code generation, turning the act of building into a trivial step. The real scarcity now lies in the human ability to discern which artifacts merit existence—what the author calls taste. This judgment cannot be automated, measured, or outsourced, making it the sole enduring craft for software engineers in the age of generative AI.