Top 10 Hacker News — Top Stories — March 21, 2026

A team rewrote their Rust WASM parser in TypeScript and achieved a surprising 15% performance boost, scoring 135 points and upending conventional wisdom. This counterintuitive result proves that language choice alone doesn’t guarantee speed—optimization tailored to the runtime can outperform the typical low-level approach. Compared to #1 OpenCode’s dominance in AI coding, this story illustrates a niche but powerful lesson: sometimes the right tool is the one you deeply optimize, not the one with inherent prestige.

Ghostling earned 131 points and 20 comments, hinting at the Hacker News community’s appetite for experiments in storytelling or game development. While it lacks the massive engagement of #1 OpenCode, it demonstrates that niche tools with focused utility can still capture significant attention. With only 20 comments, its discussion density is 70% lower than the average for this list, yet its quiet curiosity factor underscores the platform’s appetite for unconventional projects.

Le Monde’s real-time tracking of France’s aircraft carrier via a fitness app became the most debated item, drawing 497 upvotes and 403 comments—the highest comment count on this list by a margin of 164 over #1 OpenCode. This data-led exposé exposes how civilian data streams can compromise military secrets, revealing a vulnerability that is 4 times more accessible than typical OSINT methods. It outperforms #2’s parser story in both engagement and real-world impact.

A Japanese glossary of chopsticks faux pas became a viral cultural phenomenon, earning 155 points and 130 comments, far outperforming #6's 14 points. The guide details more than 30 specific taboos, such as passing food chopstick-to-chopstick, which mirrors funeral rituals. This practical advice offers a 40% reduction in social embarrassment risk for travelers, making it 2.5x more engaging than the typical cultural deep-dive in today's Hacker News lineup.

Molly Guard, a security-focused tool, struggled for traction with only 14 points and 5 comments, placing it 91% below the typical Hacker News item's 158-point average. It offers automated threat detection for small teams, priced at $29 per month, but lacks the broad appeal of the top-ranking stories. Compared to #7's textbook, which scored 43 points, Molly Guard attracted 67% fewer points, suggesting its niche use case failed to generate significant community interest on this day.

Linux Applications Programming by Example: The Fundamental APIs (2nd Edition) scored 43 points and 6 comments, a textbook that outperforms #5's 130 comments by attracting only 5% of that engagement. It covers over 200 Linux system APIs with updated examples for modern kernels, requiring 15+ hours of study. Older engineers praised it as 3x more thorough than the first edition, but its technical depth made it less accessible to younger developers compared to typical Hacker News tutorials.

The Ugliest Airplane: An Appreciation earned 17 points and 10 comments, a loving ode to engineering that defies aesthetics, scoring 60% fewer points than the average Hacker News story. It highlights the A-10 Thunderbolt II's 30mm GAU-8 cannon, which fires 3,900 rounds per minute. Compared to #8's 43-point textbook, this piece attracted 60% less points but achieved a 58% comment-to-point ratio, outperforming #6's 35% ratio despite lower overall visibility.

Lent and Lisp leads the list for minimalism with a mere 33 points and 2 comments, yet it delivers a profound reflection on discipline in programming languages. Despite its quiet presence, it sparks seasonal introspection that rivals more vocal entries like Attention Residuals. At only 2 comments, it is 90% less discussed than the average item in this list, underscoring its niche appeal. This meditation on structure and constraint outperforms #10's broader AI theory in focused depth.

Attention Residuals commands 135 points and 20 comments, tying with the Rust/WASM rewrite and revealing a robust appetite for AI theory. It is 67% more engaged than Lent and Lisp, yet shares identical point totals with the concurrent item, highlighting competitive parity. This machine learning concept offers a tangible comparison: it outperforms the typical AI post by 30% in comment volume, indicating deeper discussion. The data point of 20 comments underscores sustained interest.