January 6, 2026
Forecast: Drama
Building a rain predictor on a C64 with 1985's "XPER," expert system software
Coin Flip vs Commodore 64: The Retro Rain Showdown
TLDR: A retro experiment puts a 1985 expert system on a Commodore 64 up against human intuition and a coin to predict rain. The community is split between joyful nostalgia and sharp skepticism, turning the comments into a showdown over whether old-school rule-based “AI” can beat luck—and why that matters today.
The community is howling as a 40-year-old home computer tries to forecast rain like a tiny weather wizard. Using 1985’s XPER—an early “expert system” (think: a super rulebook turned into software)—author Christopher Drum pits a Commodore 64 (C64) against his gut and, yes, a coin flip. Nostalgia fans are cheering the yellow-screen glory, while skeptics channel Gary Kildall’s famous side-eye, calling it AI theater rather than actual intelligence. The strongest opinion? That this is deliciously pointless and somehow the most honest AI experiment of the year.
Drama arrives with the classic debate: is this real AI (artificial intelligence) or just a “competent system” doing if-this-then-that tricks? Dreyfus’s old claim that rules can only get you “85%” of expertise fuels the comments—some argue that’s all you need for rain, others say intuition still wins (or at least the coin does). Jokes fly fast: “Mushroom weatherman” memes nod to XPER’s fungi roots, “64K brain vs storm cloud” is the new banner, and weather nerds mock the hype with “no model beats an umbrella.” Fans love the retro hustle; haters love the reality check. And in the middle, everyone’s refreshing to see whether the coin dunks on the computer again.
Key Points
- •The article proposes using 1984’s XPER expert-system software on a Commodore 64 to build a rain-prediction tool.
- •Historical framing includes 1984–1985 Computer Chronicles segments with Gary Kildall, Stewart Chiefet, and Edward Feigenbaum, alongside Hubert Dreyfus’s critique of rule-based AI.
- •XPER was created by Dr. Jacques Lebbe to encode mushroom identification knowledge for use on modest microcomputer hardware.
- •A timeline shows XPER’s place in AI history and notes its recommendation for France’s 'Computing for All' initiative, pricing for C64/Apple IIe/c/DOS, and unverified mentions of Minitel and Apricot.
- •Successors Xper2 (2010, Java) and Xper3 (2013, web) added features such as image attachments, metadata, collaboration, and improved analysis tools.