Thesis 47

The practice of technological development is tending to become more decentralized.

Sometime in early 2002in an agnès b. store in Shibuya, of all placesI heard the full-throated sound of the new century for the first time. The track thumping from the store's sound system bent the familiar surge of Iggy and the Stooges' "No Fun" to the insistent beat of "Push It" by Salt'n'Pepa, and it was just as the old commercials promised: These truly were two great tastes that tasted great together. Rarely, in fact, had any music sent such an instant thrill of glee and profound rightness through me. The next track smashed the velvet Underground classic "I'm Waiting For The Man" into some gormless eighties hit, and that was almost as tasty; I found myself literally pogo-dancing around the store.

I had caught the mashup virus. A mashup is just about what it sounds like: the result of a DJ taking two unrelated songs andby speeding, slowing, or otherwise manipulating one or both of themhybridizing them into something entirely new. Anyone can do it, really, but the genius of a truly clever mashup is finding some note of deep complementarity in two source texts that seemingly could not possibly have less to do with one another. After all, until this particular gang of provocateursa Belgian duo calling themselves 2 Many DJscame along, who ever would have thought that a prime slab of Motor City protopunk circa 1969 would have worked so well against a sassily righteous hip-hop single of the mid-1980s?

I was hooked, all right. What I didn't understand at the time, though, was that I had also been given a first glimpse of one of the most important ideas to hit software development since object-oriented programming achieved widespread acceptance in the mid-1980s. The cultural logic of the mashup, in which amateurs pick up pieces already at hand and plug them into each other in unexpected and interesting ways, turns out to be perfectly suited to an age of open and distributed computational resourcesthe "small pieces loosely joined" that make up the contemporary World Wide Web, in essayist David Weinberger's evocative phrasing.

In the case of the Web, the ingredients of a mashup are not songs, but services provided by sites such as Google and yahoo! and the community site Craigslist, each of which generates enormous quantities of data on a daily basis. Services like these have at their core an extensive database that is more or less richly tagged with metadatainformation about information, such as where and when a picture was taken, or the ZIP code of an apartment listing. When millions of pieces of such self-describing data are made availabletossed on the table like so many Lego bricks, as it wereit's easy for third-party applications to pick them up and plug them into one another.

And so we see mashups from HousingMaps, which combines apartment listings from Craigslist with Google Maps to produce a searchable map of available rentals, to Stamen Design's Mappr, which uses metadata associated with images drawn from the Flickr photo-sharing service to plot them on a map of the United States. There's even a free application called Ning that lets anyone build a custom mashup, employing what might be called the Chinese menu approach: choose an application from Column A, another from Column B, and so on. It is simultaneously a profound demystification of the development process and a kit of parts that lets people without any programming experience to speak of build local, small-scale applications perfectly tuned to their specific needs.

What does any of this have to do with everyware? It suggests that we're about to experience a significant and unprecedented growth in the number of nonspecialists empowered to develop homebrew applications. It's crucial that the tools exist that allow us to do so, but still more important is a cultural context that not merely permits but encourages us to experimentand that is just what the sense of ferment around mashups provides us.

Especially when combined with the revolution in conceptions of intellectual property that was originally sparked by the open-source and free software movements, we have everything necessary to democratize the development of information technology. What was not so very long ago a matter of a few thousand corporate and academic research centers will explode to encompass tens or even hundreds of millions of independent, unaffiliated developers scattered across the globe.

As a result, everyware is not going to be something simply vended to a passive audience by the likes of Intel and Samsung: What tools such as Ning tell us is that there will be millions of homebrew designer/makers developing their own modules of functionality, each of which will bear the hooks that allow it to be plugged into others.

Of course, not everyone will be interested in becoming a developer: The far greater proportion of people will continue to be involved with information technology primarily as users and consumers. And for the time being, anyway, the sheer complexity of ubiquitous systems will mitigate the otherwise strong turn toward amateurism that has characterized recent software development. But over the longer term, the centrifugal trend will be irresistible. The practice of technological development itself will become decisively decentralized, in a way that hasn't been true for at least a century.