jon_kleinberg's picture
Tisch University Professor of Computer Science, Cornell University

When Rio de Janeiro was announced as the site of the 2016 Summer Olympics, I was on the phone with colleagues, talking about some ideas for how to track breaking news on the Internet. Curious to see how reactions to the announcement were playing out, we went onto the Web to take a look, pushing our way like tourists into the midst of a celebration that was already well underway. The sense that we were surrounded by crowds was not entirely in our imaginations: over a thousand tweets per minute about Rio were appearing on Twitter; Wikipedians were posting continuous updates to their "2016 Summer Olympics" page; and political blogs were filled with active conversations about the lobbying of world leaders on behalf of different cities.

This is the shape that current events take on-line, and there is something more going on here than simple volume. Until recently, information about an event like this would have been disseminated according to a top-down structure, consisting of an editorially assembled sampling of summaries of the official announcement, reports of selected reactions, and stories of crowds gathering at the scene. But now the information emerges bottom-up, converging in tiny pieces from all directions: the crowd itself speaks, in a million distinct voices — a deluge of different perspectives.

The Web hasn't always looked this way. When I first used an Internet search engine in the early 1990s, I imagined myself dipping into a vast, universal library, a museum vault filled with accumulated knowledge. The fact that I shared this museum vault with other visitors was something that I knew in principle, but could not directly perceive — we had the tools to engage with the information but not with one another, and so we all passed invisibly by each other.

When I go on-line today, all those rooms and hallways are teeming, and I can see it. What strikes me is the human texture of the information — the visible conversations, the spikes and bursts of text, the controlled graffiti of tagging and commenting. I've come to appreciate the way the event and the crowd in fact live in symbiosis, each dependent on the other — the people all talking at once about the event, but the event only fully comprehensible as the sum total of the human reaction to it. The construction feels literary in its complexity — a scene as though described by an omniscient narrator, jumping between different points of view, except that here all these voices belong to real, living beings, and there's no master narrative coordinating them. The cacophony might make sense, and it might not.

But the complexity does not just arise from all the human voices — it is accentuated by the fact that the online world is one where human beings and computational creations commingle. You bump into these computational artifacts like strange characters in a Carrollian Wonderland. There is the giant creature who has memorized everything ever written, and will repeat excerpts back to you (mainly out of context) in response to your questions. There are the diaphanous forms, barely visible at the right-hand edge of your field of vision, who listen mutely as you cancel meetings and talk about staying home in bed, and then mysteriously begin slipping you ads for cough medicine and pain relievers. And even more exotic characters are on the way; a whole industry works tirelessly to develop them.

The ads for cough medicine are important, and not just because they're part of what pays for the whole operation. They should continuously remind you that you're part of the giant crowd as well, that everything you do is feeding into a global conversation that is not only visible but recorded. I try to reflect on what behavioral targeting algorithms must think of me — what the mosaic of my actions must look like when everything is taken into account, and which pieces of that mosaic would have been better left off the table.

The complexity of the online world means that when I use the Internet today, even for the most mundane of purposes, I find myself drawing on skills that I first learned in doing research — evaluating many different observations and interpretations of the same events; asking how people's underlying perspectives, tools, and ways of behaving have served to shape their interpretations; and reflecting on my own decisions as part of this process. Think about the cognitive demands this activity involves — once the domain of scholarship, it is now something that the Internet requires from us on a daily basis. It suggests that in addition to "computer literacy," an old pursuit where we teach novices how to use computing technology in a purely operational sense, we need to be conveying the much more complex skill of "information literacy" at very young ages: how to reason about the swirl of perspectives you find when you consume information on-line, how to understand and harness the computational forces that shape this information, and how to reason about the subtle consequences of your own actions on the Internet.

Finally, the Internet has changed how I think professionally, as a computer scientist. In the thirteen years since I finished graduate school, the Internet has steadily and incontrovertibly advanced the argument that computer science is not just about technology but about human beings as well — about the power of human beings to collectively create knowledge and engage in self-expression on a global scale. This has been a thrilling development, and one that points to a new phase in our understanding of what people and technology can accomplish together, and about the world we've grown to jointly inhabit.