Bilingual

Matt’s Script Archive: The Scripts That Reshaped The Web
17 hours ago
- In the mid-1990s, high school student Matt Wright created simple web tools like guestbooks, contact forms, and the popular WWWBoard forum, hosted on Matt's Script Archive.
- His scripts were widely adopted by non-developers but were riddled with security flaws, including a critical 10.0 CVE for the textcounter tool that allowed remote code execution as root.
- The nms project was launched to provide drop-in replacements for Matt's insecure scripts, but it hasn't been updated in about 20 years due to changes in web development and security practices.
- The tension between easy-to-use tools and security mirrors modern 'vibe coding'; Matt's scripts were both a democratizing force and a security risk.
- The domain worldwidemart.com (formerly hosting Matt's Script Archive) was revived by a fan to preserve the historical legacy of the scripts, focusing on education rather than spam.
- Wright himself later acknowledged his code was outdated and recommended nms instead, illustrating the gap between novice and professional development perspectives.
Pluralistic: AI and amateurism (15 Jun 2026)
17 hours ago
- The author's lifelong involvement with technology, from a cardboard computer to digital rights activism, shapes his perspective on AI.
- Cyberpunk's dualism—technology as both liberating and oppressive—is key to understanding AI's potential and pitfalls.
- Tools like Hypercard and Scratch empower non-programmers to create software that meets their specific needs, a principle extended by 'vibe coding' for personal projects.
- Vibe coding is valuable for personal automation but is misrepresented by industry hype as a replacement for professional software development.
- The AI industry focuses on centralization and homogenization, while local models and amateur tinkering preserve idiosyncrasy and user control.
<antirez>
17 hours ago
- Automatic Programming is the process of writing software with AI assistance where the human actively guides the process using intuition, design, and continuous steering.
- Vibe coding is a passive approach where a user gives vague descriptions and the AI produces code without meaningful human involvement or understanding.
- Code generated through automatic programming belongs to the human producer, as pre-training data is a collective human gift and the human's vision is essential.
- The example of Redis shows that software value comes from ideas and vision, not just technical novelty, illustrating that programming is now automatic but vision is not.
Tool: LLM cliché highlighter
17 hours ago
- The LLM cliché highlighter is a tool that detects and highlights common clichéd phrases in LLM-generated text.
- It was created by the user (via Fable 5 vibe code) due to frustration with cliché-ridden articles.
- The tool features toggleable pattern detection and context-aware sentence highlighting for ten common patterns.
- The text also mentions recent articles on OpenAI, a fireside chat with the Claude Code team, and Kimi K3.
Apple won't let me show my app
4 days ago
- Apple's App Store is experiencing high submission volume due to 'vibe coding', causing delays in app review, even for major apps like X.
- Non-game apps have surpassed games as the top revenue source on the App Store for the first time, likely due to easier CRUD app development.
- The app review team often uses automated rejection messages with minimal detail, making it hard for developers to communicate their perspective.
- The developer of Wiseday faced a rejection because screenshots supposedly didn't show the actual app in use, despite them illustrating the core concept (a paper page).
- The developer argues that Apple's two guidelines—showing the actual app in use and highlighting the core concept—are contradictory for Wiseday, as the core concept is a paper page previewed in-app.
- The developer responded to the rejection by explaining the technical details and adding a traditional device-wrapped screenshot to comply with the majority rule.
How Do We Stop Vibe Coding?
4 days ago
- Coding agents rapidly gained popularity among software developers after Claude Code's surge in late 2025, with figures like Andrej Karpathy noting a shift away from manual coding since December 2025.
- Grady Booch views this as the third golden age of software engineering, where AI agents abstract code itself, similar to how compilers abstracted the machine.
- The core problem with 'vibe coding' is the loss of understanding, awareness of dead code, and the ability to reason about or explain architectures, leading to unreliable outputs.
- Current solutions like markdown specs, skills, Spec Kit, OpenSpec, Kiro, and test-driven development fail to address trust, offering only prompt engineering without deterministic enforcement.
- Promising approaches like CodeSpeak and Scryer focus on trust by compiling specs into code or creating a shared model between developer and agent for transparency and drift detection.
- The future likely involves moving beyond manual coding for routine tasks, with agents guided by structured, auditable intent rather than vague prompts.

vibe coding

Matt’s Script Archive: The Scripts That Reshaped The Web

Pluralistic: AI and amateurism (15 Jun 2026)

<antirez>

Tool: LLM cliché highlighter

Apple won't let me show my app

How Do We Stop Vibe Coding?