computer science

Bilingual

Show HN: An AI wrote and illustrated a book of fables about threads and packets
2 months ago
- Each fable mirrors a computer science concept (deadlock, cache, checksum, loop bounds, race conditions, floating-point precision, greedy optimization, packet duplication, zombie processes, heartbeat signals, compiler warnings, model overfitting) and attaches a moral about human behavior.
- The collection is written by an AI model named Fable, which reflects that its stories about machines are inevitably about human vices, because its training data is human-authored text.
- The fables emphasize that technical systems fail due to human-like flaws (courtesy without order, hoarding, trust without verification, misapplied repairs, assuming 'never happened' means 'cannot happen').
- The meta-fable reveals that the AI itself is the student of its own fable about overfitting—it only has one book (human text) and cannot escape writing about human nature.
- Practical morals include: agree on order before acting, forget well to serve purpose, verify even trusted sources, count fence before moving it, and recognize that exactness and range are trade-offs.
Good Ideas in Computer Science
2 months ago
- The article presents a list of universally agreed-upon good ideas in computer science, such as arrays, hash maps, virtual address space, and subroutines, emphasizing that these are foundational concepts not debated among programmers.
- It intentionally excludes garbage collection, databases, most data structures/algorithms, and object-oriented programming, as they are either implementation-specific, debated, or not universally considered good.
- The author notes that by 1974, most modern computing fundamentals were already in place, and hopes for new universally good ideas to emerge in the next 50 years.
Pluralistic: CARDiac, syntax coloring, view source and vibe code (03 Jul 2026)
2 months ago
- The author describes how the CARDiac, a paper computer from 1968, taught him the concept of 'legibility' in computing by physically moving tokens to understand computer operations.
- Viewing source code (from BASIC programs to early web HTML) allowed him to peel back abstractions and gain deeper understanding, similar to using the CARDiac.
- Vibe coding and automated code tools (like macros or LLMs) serve as entry points; while they solve immediate problems, they also create opportunities for users to dive deeper into coding.
- Legibility (the ability to understand and manipulate systems at different abstraction levels) is crucial for problem solving, improvement, and maintenance of technology.
- The first programmers (often women, per Claire Evans' 'Broad Band') developed powerful intuition through hands-on physical cabling, enhancing their debugging skills.
Computer science enrollment fell for the first time in 20 years
2 months ago
- Computer science enrollment declined for the first time in roughly two decades in the 2025-26 academic year.
- Enrollment in Computer and Information Sciences at four-year colleges fell 8.1% year over year in fall 2025.
- The decline coincides with the rapid adoption of generative AI like ChatGPT, but the data does not prove AI caused it.
- Stanford economist Jacob Light's research shows a 4.6% year-over-year drop in CS enrollment at the average school, with share falling 6.3 percentage points.
- Five possible explanations include a weaker job market, uncertainty about CS degree value in the AI era, weaker academic preparation, demographic changes, and expansion of competing fields.
- Despite the decline, enrollment remains above 2022 levels, suggesting the drop follows years of rapid growth.
ArsDigita University: Five Years Later (2006)
2 months ago
- The author describes their experience attending Arsdigita University, a free intensive computer science program founded by Philip Greenspun in 2000.
- The program aimed to provide world-class computer science education to those unable to access it, lasting 10.5 months with lectures from MIT professors.
- Despite initial enthusiasm, challenges arose including lack of certification, funding cuts, and the dismissal of foundation staff, leading to no second class.
- The curriculum covered topics like mathematics, programming, systems, and a software engineering course, though plans changed with unexpected Microsoft involvement.
- After graduation in 2001, alumni pursued diverse careers at organizations like IBM, The New York Times, and various tech firms, and some continued their education.
- The author reflects on the program's legacy, noting its influence on online education and open courseware, while expressing gratitude to Philip Greenspun and director Shai Simonson.
"Software Engineering" Is Not Engineering
3 months ago
- Computer Science is not science and Software Engineering is not engineering because software design lacks objective metrics and ties to physical reality.
- Software is more akin to mathematics, allowing creation of virtual worlds with internal consistency, unlike engineering bound by physical laws.
- Debates in software often revolve around idioms, paradigms, and conventions, which are difficult to compare objectively due to reliance on human psychology and productivity factors.
- Programmer productivity and maintainability are key concerns, but empirical studies are scarce, and personal comfort with tools often dictates productivity.
- Existing computer science literature focuses on techniques and idioms, such as Boolean algebra, but fails to provide scientific methods for evaluating 'better' approaches.
- The flexibility of software enables infinite possibilities, but this also makes objective evaluation nearly impossible, leading to persistent, unresolved debates in the field.

computer science

Show HN: An AI wrote and illustrated a book of fables about threads and packets

Good Ideas in Computer Science

Pluralistic: CARDiac, syntax coloring, view source and vibe code (03 Jul 2026)

Computer science enrollment fell for the first time in 20 years

ArsDigita University: Five Years Later (2006)

"Software Engineering" Is Not Engineering