If there is one thing I enjoy about the web, it's the way that primary historical materials are as close to hand as a Gawker blog post or a Amazon.com. Getting your hands on real, printed photographs from the 1910s takes work, even if you happen to have a university affiliation.
Online, as part of our daily labor here on The Atlantic, I often find myself at the Library of Congress searching through hundreds of thousands of photographs of all kinds of things. At a time when algorithms are supposed to be reducing serendipity to the opposite of a chance encounter, I find the blunt search tools at the LOC constantly spit out wonderfully unexpected things.
For Alexander Furnas' story yesterday about power, privacy, and data tracking, I wanted to find a photograph of a bunch of dolls, so I searched for "doll shop." Scrolling down the list, I didn't find what I wanted, but one title for a group of photos caught my eye: "Old men making toys in a shop maintained for their benefit, apparently by society women." The record told me George Grantham Bain made these pictures in 1915. The extended description read, "Photographs show men cutting animals and dolls from wood. Women purchasing Christmas gifts. Also, teddy bear factory." There is no more information attached to the record, but who really needs more than TEDDY BEAR FACTORY, really.
Take a look at these photographs. They tell you something about how and where toys were once made. But they also tell you something mustaches and hairstyles, about the way people tucked in their blouses, and what a teddy bear looked like in 1915. You can see how many humans worked in a teddy bear factory and how they held themselves when someone walked onto the floor with a camera.
What I'm trying to point out is this: Algorithms may take the brownian motion out of discovering stuff, but it's not as if adjacency goes away. Serendipity, for me, is nothing more than an openness to seeing.