February 9, 2026
Built on vibes and error messages
Nobody knows how the whole system works
We’re rolling out black boxes and arguing who’s liable
TLDR: Experts warn modern systems are built without anyone fully understanding them, echoing an old MIT story about phones. Comments clash over AI-written code, accountability when things fail, and whether old-school fundamentals still keep us safe—raising big questions about trust in the tech we use every day.
LinkedIn is suddenly the place where tech elders worry out loud, and this week it’s all about building stuff we don’t fully understand. Lorin Hochstein rounded up posts from Simon Wardley, Adam Jacob, and Bruce Perens, plus a throwback from MIT’s Louis Bucciarelli asking, “Does anyone know how their telephone works?” The mood: we’re stacking Lego towers in the dark, praying they don’t topple.
The comments turned it into a full-on courtroom drama. youarentrightjr shrugs “true,” but warns the old rule—someone knows each piece—is eroding. mamp jumps in with the AI bombshell: if bots write the code, maybe nobody knows the system at all. virgilp waves the basics flag: lose foundational understanding, lose control. whytaka demands accountability: fine to work at a high level, as long as you’re not blamed when it breaks. Then bjt slams the brakes, saying tried-and-true computing fundamentals are solid, unlike today’s large language model (LLM) code that’s “leaky” and constantly reworked. Cue memes: “Ship it and pray,” “duct-tape rockets,” and “I for one welcome our black-box overlords.” The fight is simple, loud, and very online: Are we okay living with mystery machines—or are we quietly building a house of cards?
Key Points
- •The post highlights a shift from Twitter to LinkedIn for industry discussions on complex systems.
- •It references Simon Wardley’s LinkedIn post about building systems without fully understanding them, with responses by Adam Jacob and Bruce Perens.
- •Louis Bucciarelli’s 1994 book is quoted to question what it means to ‘know’ how a telephone works across multiple layers.
- •Bucciarelli’s examples include physical mechanisms, routing heuristics, signal processing algorithms, satellite links, multi-operator networks, field operations, and regulation.
- •The article ties historical reflections on distributed understanding to current conversations about complexity and system knowledge.