11 hours ago
- The project of software is complete; the era of new software categories has ended, and current efforts are refinements rather than innovations.
- Despite the democratization of coding via LLMs ('vibe coding'), most new software is redundant or trivial, not addressing genuine unmet needs.
- The adjacent possible of software is saturated; new apps like habit trackers are terminal leaves, not branches that open new frontiers.
- LLMs are powerful at manipulating symbols but cannot solve physical-world problems (e.g., energy, disease, transportation) that require atoms, not bits.
- Future technological progress must come from material science and engineering, not from more layers of software abstraction.