January 29, 2004
Google Pipe Binary Atomization (Objects, not Products)

Several years ago, Marc Smith at Microsoft Research was talking about a future in which all products in the world had a kind of economic and even social transparency to them, because their bar codes could trigger cascades of comparative data -- if you knew how to pull it from the air.

Smith is sharp like tacks and teeth are sharp, but I confess that at the time, I found this vision kind of naive. It may have been the ways that Smith was talking about it -- that we would pull indexed information on corporate social policies, and make our buying decisions based on that. Not because I don't want it, but because I don't believe in America like that.

The imagery that Smith brought to mind was made concrete for me in James Patten's clever but spoofish Corporate Fallout Detector. Idealistic, but not so pragmatic; even if you could do the same thing with a PDA, I see few precedents to suggest that anyone would want to. The technology itself seemed plausible, but I couldn't recognize any social future that resembled the story Smith was telling. Historically, in the US -- as opposed to Europe -- most social values are voiced through political choices (voting) as opposed to consumer choices (buying).

Beyond the social indices, you could get comparative price information, but look at where Amazon is today and where all the "comparebot" technologies went by comparison. There's a million contextual purchasing factors, like convenience, environment, reliability, and reputation; all of these determine the value of products in those moments we make buying decisions. The comparative index of other available prices is only one piece of data, often the least important. Think about people who pay $10,000 more for the Escalade than the identically constructed Tahoe, just because the Escalade has this on it. It's America, we buy brands and experiences, not products. We don't need a GPS receiver to tell us which way the wind blows.

I'm changing my mind, though, in part because I discovered Google's newest suite of tools the other day. "Search by Number" provides tools to use numbers to track UPS, FedEx and USPS packages, but also auto VIN codes, FAA airplane registry information, FCC equipment, UPC codes and even patents. It's predictably extraordinary and obvious, but it feels as if crosses two thresholds.

First, using Google to track packages, products, vehicles and airplanes represents the closest points of tangency between google and the tactile world. Search engines in their original conception are echoes of echoes, indexing a world of homepages and blogs that is already parallel to real life. But using them to track objects SM+L means that consumer technologies are indexing the world, not the web.


Second, related, is simply the implosive sense of collapsing everything -- patents and airplanes and every consumer product -- down to binary form. It's a reverse Matrix, every concrete thing in the world rendered unto data, a kind of binary atomization. Every object in the world with unique digital identification, all of it now moving through a single Google conduit. What’s remarkable about this stream of data is that it’s so thin, somehow singular.

I think that when I found it difficult to imagine Smith's future, part of what I lacked vision for was some kind of aggregate conduit (outside of its most negative possibilities). So now there's a google index for all the objects in the world, the ones in the stores, mail, air and transit. Google knows what consumer goods corresponds to what UPC codes. For example, Google might know that the #036632006226 I’m eating right now is 8 ounces, but it doesn’t know which 8 ounces; to these eyes, every Blueberry yogurt looks the same.

But that will probably change, soon, because beyond the Google number-world, there’s two other technologies in play right now. I just discovered ePC codes, which are like UPC codes, but each one is unique:

“There are several key differences between an EPC and a bar code. First, the EPC is designed to provide a unique serial number for every item in the system. By contrast, bar codes only identify groups of products. So, all cans of Diet Coke have the same bar code more or less. Under EPC, every can of Coke would have a one-of-a-kind identifier.”

The second technology, RFIDs, are media-visible, in part because privacy advocates have thrown the appropriate flag. What RFIDs do is add geography to the data cloud. There’s nothing abstract about it, even as the real plans for Gillette’s RFID-embeds calcify into myth. There’s nothing abstract about it: products will know where they are. Combine RFID and ePC, and it’s not the geography of products, it’s the geography of objects.

So now it’s not that a razor knows where it is, it’s that this razor knows where it is. This thought is recent but not new, it’s a substantial part of what Howard Rheingold addresses in Smart Mobs. The thing that’s new is the idea of a civilian technology – namely, Google – aggregating and linking all that data at street level.

Marc Smith’s example, the one I couldn’t envision at the time, was of pervasive access to product information. “Google numbers” resolves that, more or less, since the same pipe that brings you news also resolves codes. But it goes far further: with ePCs, you’re not googling the product, you’re googling the object. Where it is, where it might have been, its status, perhaps. In addition to all the contextual information surrounding that object, e.g., the web.

As of today there’s one channel that’s tracking where FedEx packages and airplanes go. That same channel will soon include all the data on all the other objects: razors, yogurt, whatever. Google becomes the network to which everything in the world is broadcasting, and which everyone is watching. Like cloning, or dumping robots on oblivious and distant planets, if we can do this, we probably will.

Or: It’s not a search engine. It’s an atlas.

Posted by kevin slavin at January 29, 2004 02:17 AM

As a postscript to this (and there should be more), see Roland Piquepaille's "Googling Our Genome."

"[For example,) a bioinformatics program running on a PC could easily check our genomes for all genes associated with [a particular disorder.] Regular software updates downloaded from the internet -- like those for anti-virus programs -- would keep our search software abreast of the latest medical research."

Posted by: Kevin Slavin on April 23, 2004 05:35 PM

kevin, i don't know how i missed it the first time, but you hit 1,000 words in that post. which is a saturation record, as far as i can tell.

Posted by: elia on April 25, 2004 03:06 PM

It does my heart good to know that even if no one's reading my words, at least Elia is out there counting them.

Posted by: Kevin Slavin on April 26, 2004 12:43 AM
Post a comment

Email Address:


Remember info?