Simple Made Inevitable: How LLMs Are Changing the Economics of Programming Language Choice
Key Takeaways
- ▸LLMs eliminate the traditional "learning curve" barrier, making developer familiarity less important than language simplicity when choosing programming languages
- ▸Anthropic's Claude demonstrated ability to quickly absorb complex Clojure frameworks and write fluent code, suggesting AI coding agents evaluate languages by their abstractions rather than syntax familiarity
- ▸The distinction between "simple" (objectively unentangled) and "easy" (familiar) languages becomes economically critical when AI agents are primary code producers
Summary
A long-form essay by Felix Barbalet argues that AI coding agents are fundamentally changing how organizations should evaluate programming languages, making simplicity more valuable than familiarity. Drawing on experiences with Clojure at Qantas and referencing Fred Brooks' distinction between essential and accidental complexity, Barbalet contends that LLMs eliminate the traditional "learning curve" barrier that kept developers from adopting more principled but less familiar languages. The piece cites examples from Nathan Marz, who used Claude to build complex distributed systems in Clojure, and Wes McKinney, creator of pandas, who now writes extensively in Go despite never having learned it manually.
The core thesis challenges conventional wisdom about language selection: for fifteen years, Clojure and similar "simple" languages faced adoption resistance because they weren't "easy" — they lacked familiar syntax, large talent pools, and extensive Stack Overflow coverage. But LLMs don't experience learning curves or need weekend tutorials. They evaluate languages purely on their intrinsic abstractions and freedom from "accidental complexity" — the overhead imposed by tools rather than problems themselves. This shift, Barbalet argues, makes developer familiarity a diminishing selection criterion.
The article positions this as an economic inflection point rather than merely a technical one. When AI agents can fluently write in any language after minimal exposure, the calculus shifts from "what can we hire for?" to "what minimizes the complexity our AI collaborators must navigate?" Languages designed around immutable data, pure functions, and compositional abstractions — traditionally dismissed as too academic or niche — may suddenly offer superior economics in LLM-assisted development environments.
- Languages designed to minimize accidental complexity through immutable data and compositional abstractions may offer better economics in AI-assisted development than conventionally popular languages
Editorial Opinion
This represents a genuinely provocative thesis about AI's second-order effects on software engineering. While the author's Clojure advocacy is transparent, the underlying economic argument deserves serious consideration: if coding agents become primary producers and human developers primary reviewers, then optimizing for machine reasoning clarity over human syntax comfort isn't contrarian — it's rational capital allocation. The industry may be on the cusp of relitigating decades of language design tradeoffs through an entirely new lens.
