Loading page
Learn
Reverse image search, niche indexes, and search engines built on a premise other than popularity. What the web looks like when ranked differently.
17 sites
Searches the web for actual sound files by keyword, in a layout untouched since roughly 2003. Still the fastest route to a duck quack.
Grey text on beige, no images, and the best comment threads anywhere about how technical things actually work. Search covers twenty years.
Full-text search across every Hacker News story and comment, filterable by date and points. Find the thread where a famous project first appeared.
A link aggregator about computing that anyone can read: smaller, quieter and more carefully tagged than the big tech news sites.
An index deliberately built to favour text-heavy non-commercial pages. Ranking that punishes ad load instead of rewarding it.
Live map of which McDonald's ice cream machines are broken, built by repeatedly adding a sundae to a cart. Petty, rigorous, brilliant.
Search the web with the top million most popular sites removed. What is left is the internet nobody bothers optimising for.
A search engine that crawls the web with its own bot rather than renting results from the giants, and does not track who is searching.
A non-profit, open-source search engine with its own index, curated by volunteers. Results feel nothing like the big engines, which is the point.
Searches around a billion openly licensed images and audio files, with the licence terms attached to every single result.
Reverse image search tuned for illustration: hand it a cropped screenshot and it names the artist and the original post.
Indexes only independent personal websites whose owners submitted them. No content farms, because there is nothing there to farm.
Search engine for internet-connected devices rather than pages: webcams, routers and industrial kit nobody meant to publish.
Non-commercial web search indexing only pages carrying no ads or trackers at all. A very small and very different internet.
Reverse image search that finds the oldest copy of a picture rather than visually similar ones. The tool for tracing where a photo began.
Search engine for hand-made pages in the spirit of the early web, with a surprise-me button that behaves exactly like 1998.
Widely reckoned the strongest reverse image search for faces, places and heavily cropped photos. Worth trying when TinEye finds nothing.
Every entry in Detour is hand-sorted rather than scraped: filed under one category, given one to three moods, and placed in a time band for how quickly it pays off. The practical facts — cost, sign-up, sound, device and whether it is safe at a desk — are recorded per site and shown on every card, so you can rule things out before you click.
17 sites, each hand-sorted by category, mood and how long it takes to pay off.
17 of the 17 are free and need no account. None of them are paid.
12 of 17 work on a phone; the rest are desktop only. 17 are safe for work and 16 work with the sound off.
One button, one detour. Press it as often as you like.