Good Taste the Only Real Moat Left
AI and LLMs have made competent output cheap, forcing a re-evaluation of what truly creates value in the tech landscape. This article argues that human 'taste'—the ability to discern, reject, and precisely articulate what's wrong with generic AI outputs—emerges as a critical moat. Hacker News debates whether 'taste' is truly the new bottleneck or if traditional moats like distribution and execution still reign supreme.
The Lowdown
The advent of AI and Large Language Models (LLMs) has fundamentally altered the landscape of creative and technical work, making 'competent' output readily achievable. This article posits that in this new era of abundant average, human 'taste'—defined as the ability to make distinctions under uncertainty and precisely articulate why something is or isn't effective—becomes the crucial differentiator. It argues that while AI excels at pattern matching and generating plausible, often generic, results, it lacks the context, stakes, and authorship necessary for truly impactful work.
- AI flattens the middle by easily producing statistically plausible, yet often generic, outputs (e.g., boilerplate landing pages, bland product copy).
- Human taste is essential for moving beyond 'vibe' to 'diagnosis,' enabling precise critique and rejection of generic or context-blind AI-generated content.
- The real value in the age of AI shifts from generation to judgment, specifically the ability to discern what is truly specific, useful, and aligned with real-world constraints.
- The author suggests a practical loop for training taste: generate many AI versions, diagnose their failings precisely, rewrite with hard constraints, and ship the result to learn from real consequences.
- However, taste alone is not sufficient; humans must retain authorship, hold real-world stakes, work with genuinely new ideas, and make directional decisions that AI cannot.
- Builders should use AI to explore the design space and understand existing patterns, but then apply their unique judgment, constraints, and point of view to create specific, trustworthy products that transcend mere polish.
Ultimately, the article advocates for a symbiotic relationship with AI where human judgment is sharpened, not replaced. Taste is seen as a byproduct of serious, consequence-laden work, encouraging builders to leverage AI's speed while vehemently guarding specificity, context, and the courage to create something truly distinctive.
The Gossip
Moats & Moanings: Is Taste the New King?
Commenters were sharply divided on the article's central premise that 'taste' is the last remaining moat in the age of AI. Many argued that traditional business advantages like distribution, proprietary data, iteration speed, and sheer human effort remain far more significant. Skeptics pointed out that 'mediocre slop' has always been prevalent and questioned techies' perceived 'good taste.' Conversely, proponents emphasized the author's point that precise judgment is indeed a critical skill, allowing humans to guide AI towards 'perfect' outcomes rather than just accepting average. The debate also touched upon the connection between taste, 'skin-in-the-game,' and real-world consequences.
AI's Autonomy vs. Human Authorship
A significant thread explored whether AI itself could eventually develop 'taste' or discernment, with some linking this to the concept of AGI and the 'bitter lesson.' Others highlighted the practical reality of AI's current use: it excels at generating plausible, average output, which can be strategically deployed for low-stakes situations (like 'slop' documents no one will read). This underscores the human role in providing precise direction and understanding contextual constraints, preventing the AI from merely producing generic results and ensuring genuine authorship.
Historical Echoes and the Cycle of 'Newness'
Some commenters found the article's core idea to be an age-old debate, drawing parallels to historical discussions about design vs. functionality (e.g., Steve Jobs vs. Bill Gates) or the pitfalls of outsourcing core competencies. They suggested that the perceived 'newness' of the problem might just be a contemporary re-framing of enduring challenges, particularly concerning the generation of 'mediocre slop.' Yet, others acknowledged that the current AI-driven shift in economics *feels* genuinely new and warrants re-evaluation of career value.