Project ideas from Hacker News discussions.

Show HN: I spent 3 years reverse-engineering a 40 yo stock market sim from 1986

📝 Discussion Summary (Click to expand)

1. AI‑assisted writing and the “LLM‑style” debate
Many commenters focused on whether the article was written with Claude/ChatGPT and how that affects credibility.
- “I used Claude to write a lot of it… I just wanted to share the story…” – benstopics
- “I’m sensitive to detecting LLM writing… I can tell that your natural narration style is much higher quality…” – nerdsniper
- “If you’re going to write it with an LLM, do so unapologetically…” – fallinditch

2. Technical strategy for reviving a 90‑s BASIC engine
The discussion repeatedly highlighted the FFI bridge, DLL loading, and the decision to keep the core in PowerBasic while wrapping it in a modern UI.
- “I built an FFI via event dispatching and shared memory pointers…” – benstopics
- “I still use the PB compiler… I will build a virtual machine…” – benstopics
- “The core is still in BASIC; I’ll keep it that way because Michael can work on it.” – benstopics

3. Release & publishing strategy
Commenters debated the merits of self‑publishing versus a publisher, the Early Access date, and the impact on sales.
- “I will set an Early Access release date of 1‑2 months from now…” – benstopics
- “I didn’t think I would need a publisher… but it would have been nice.” – benstopics
- “If it fails it fails; we keep the proceeds and do it our way.” – benstopics

4. Community nostalgia and game interest
Readers expressed excitement about the game’s legacy, the desire to play or mod it, and the broader appeal of spreadsheet‑style simulation games.
- “I love spreadsheet games like Terra Invicta/Paradox/Simutrans…” – jjmarr
- “I’ve played the DOS version… I want to buy it.” – vessenes
- “The game sounds like a terrific example of one.” – jjmarr

These four themes capture the bulk of the discussion: the clash over AI‑generated prose, the technical hurdles of porting legacy code, the self‑publishing path chosen, and the community’s nostalgic enthusiasm for the game.


🚀 Project Ideas

LegacyDoc AI

Summary

  • Generates human‑readable documentation, variable renaming, and call‑graph visualizations for legacy codebases (e.g., PowerBasic, old BASIC, COBOL) using LLMs and static analysis.
  • Enables developers to understand, refactor, or port legacy engines without starting from scratch.

Details

Key Value
Target Audience Legacy‑software maintainers, game remaster teams, academic researchers
Core Feature Automated code‑to‑doc pipeline: parse, infer types, rename, produce Markdown + diagrams
Tech Stack Python, tree-sitter, OpenAI/Claude API, Graphviz, VS Code extension
Difficulty Medium
Monetization Hobby

Notes

  • HN users like “I need to understand 115 KLOC PowerBasic” (benstopics) and “how to port this code” (pillefitz) would appreciate a ready‑made docs generator.
  • Sparks discussion on LLM‑assisted reverse engineering and the trade‑offs between automation and human insight.

LegacyTestGen

Summary

  • Automatically generates unit and integration tests for legacy game engines by instrumenting the code and recording state transitions.
  • Provides a CI‑friendly test harness that validates core logic (e.g., financial simulations) without manual test writing.

Details

Key Value
Target Audience Game developers, QA engineers, open‑source maintainers
Core Feature Runtime instrumentation → event capture → test case synthesis
Tech Stack C/C++ instrumentation (Frida, Dyninst), Python, pytest, Docker
Difficulty High
Monetization Revenue‑ready: subscription for enterprise use

Notes

  • Addresses “manual smoke testing” frustration (benstopics) and “need for automated testing” (pillefitz).
  • Encourages best practices in legacy code maintenance and invites community contributions to test suites.

LegacyEngine Wrapper

Summary

  • A framework that exposes a legacy engine (e.g., PowerBasic) through a modern UI layer (Electron, web, mobile) via FFI, while preserving the original logic.
  • Includes a plugin system for mods and community extensions without touching the core engine.

Details

Key Value
Target Audience Indie devs, modders, retro‑game enthusiasts
Core Feature DLL bridge, event queue adapter, UI abstraction, mod API
Tech Stack C++/Rust for bridge, Electron/React for UI, WebAssembly for web, Swift/Kotlin for mobile
Difficulty Medium
Monetization Hobby

Notes

  • Resonates with “I want to run this on macOS/Linux/mobile” (logdahl) and “modding community” (kolinko).
  • Provides a reusable pattern for other legacy games, fostering a broader retro‑gaming ecosystem.

LLM Content Detector

Summary

  • A web service that analyzes text or code to detect LLM‑generated content, offering confidence scores and style markers (e.g., overuse of “it’s not X, it’s Y”).
  • Helps communities like HN flag AI‑generated posts and maintain content quality.

Details

Key Value
Target Audience Online communities, moderators, researchers
Core Feature NLP model trained on LLM‑style vs human‑style corpora, API & UI
Tech Stack Python, HuggingFace Transformers, FastAPI, React
Difficulty Medium
Monetization Hobby

Notes

  • Directly responds to the “LLM‑style writing” frustration (grey‑area, nanc, etc.) and the need for “source‑verification” (Eisenstein).
  • Sparks debate on AI transparency and the future of content moderation.

Read Later