March 4, 2026
Please hold—elevator roast in progress
Elevator Saga: The elevator programming game (2015)
Elevator Saga ignites a showdown: humans vs chatbots
TLDR: Elevator Saga, a browser game where you program elevators, has the internet debating whether humans or chatbots run the lift better. Commenters celebrate the puzzle, roast AI misses (Opus’s “monstrosity”), and claim no bot beats every level—making this a playful stress test for modern AI hype.
An old-school browser game, Elevator Saga, is back in the spotlight, and the comments are the real ride. Players code a virtual elevator, and the crowd instantly split into camps: nostalgia nerds and AI skeptics. technothrasher calls it “the hard drive scheduling game,” while agentultra flexes their favorite TLA+ elevator exercise (that’s a mathy way to plan systems). Many laugh at the starter code that just ping-pongs between floors, like a toddler mashing buttons. The vibe? Fun, tricky, and surprisingly addictive.
Then the drama hits: can large language models (chatbots) beat all levels? withinboredom says it’s a “great test for vibe coding,” confessing it took hours and that “vibe coding… is not exactly faster,” though Opus coughed up a pass. eknkc fires back: “Opus screwed it up and created a monstrosity” on the first try. Meekro drops the hammer: no chatbot can beat every level—humans still rule this shaft. Jokes fly about elevator etiquette, button rage, and algorithm throwbacks, making GitHub feel like a retro arcade with a modern AI roast. Meanwhile, seasoned pros reminisce about class days and formal methods, turning this tiny lift sim into a battleground for bragging rights and bot-bashing. The comments are pure gold.
Key Points
- •Elevator Saga is a game for programming elevator control logic using an event-driven API.
- •The page provides an example script with init and update functions and an idle event that queues floors.
- •Performance metrics (transported, elapsed time, avg/max wait, moves) are displayed to assess solutions.
- •Help, API documentation, a wiki, and a test runner are available to support and validate implementations.
- •The game credits Magnus Wolffelt, is at version 1.6.5, and its source code is available on GitHub.