Eight years after World66 went dark, we are bringing it back as World66.ai: a free travel guide where AI is used to maintain the commons, not strip-mine it. The code is open source, the text is free to copy, there are no ads or paid links, and anyone can contribute by putting an agent to work and opening a pull request. The bet is simple: AI does not get to bury the web under infinite travel slop. If it learned from the commons, it should help rebuild one.
The shape of travel guides has always followed technology. Baedeker published his Rhein guide for fast travellers in an age when railways and steamers were making travel timetabled, organized, and available to the new middle class. A century and a half later, Tony and Maureen Wheeler travelled overland from London to Australia, buses were ploughing routes that had once meant camels, and printing had become cheap. The backpackers' bible was born.
World66 was the next turn of that wheel. If cheap travel and cheap printing made Lonely Planet possible, the internet made it possible for travellers not just to read the guide, but to write it. That was the whole idea: a travel guide to the world, written by people who had actually been there. We launched before Wikipedia, which made it feel slightly mad at the time and obvious a few years later.
For a while, that seemed to be where the web was going: a giant shared notebook, messy, useful, and kept alive by people who cared about the thing they were writing about, overseen by a benevolent Google that sent traffic to pages that deserved it and wrote small checks through AdSense to help fight the costs.
As often, this model contained the Hegelian seeds of its own destruction. Professional listicle writers replaced amateur bloggers, loud video ads replaced subtle AdSense text links, and travel bookings became too big a business for the old web innocence to survive. SEO became more important than actually having been to the place. The travel web killed Lonely Planet and then drowned in human slop.
Then AI arrived, trained on all of it: guidebooks, forums, blogs, reviews, wikis, and probably plenty of World66 too. The obvious bad outcome is easy to see: the machine eats what is left of the commons, produces more slop, and calls the result knowledge.
World66.ai is a bet on using AI to support the commons instead. A world travel guide is too large for a small team to maintain manually and too tedious for volunteers to keep current page by page. But teams of humans and agents can get the job done: humans set the standards, agents do the boring work, GitHub coordinates the changes, and the whole thing scales.
The old World66 was “the travel guide you write.” World66.ai is the travel guide your agent writes: humans still decide what matters, what belongs, and what gets merged.
So that is the experiment. World66.ai is live, the repository is open, and there is plenty to do: thin pages, stale sections, missing places, broken links, awkward structure. Pick a place you know, or a place you want to research. Fork the repo, run an agent, improve one destination, and send a pull request. A guide to the world was never going to be written by one editor anyway.
Never try to teach a pig to sing. It wastes your time and annoys the pig.