In the mid-1990s, David Brock and Sanjay Sarma, engineers at MIT, became preoccupied with a problem that has long vexed robotic scientists -- how do you get a computer to understand what's happening in the physical world around it?
The standard approach is to have the machine emulate human beings; you build a robot with "eyes," with the crude ability to marshal rays of light into a "mental image" of its surroundings. But that is an enormously complex task, one that researchers have been working for decades to perfect, and Brock and Sarma wondered whether it might be possible to give the robot a little help. Instead of having the machine look around and guess what it could see, what if objects in the room simply identified themselves to the robot? Why couldn't a book carry some sort of electronic marker to alert the robot that there was a copy of Tolstoy nearby? Why couldn't a can of Coke "know" that it was a can of Coke, or a bar of soap know that it gets slippery when wet? "That way," says Sarma, "the robot could just ask the item what it was and then look up a database to see how to pick it up."
Life would certainly be much easier for robots if every item in the world -- every book, every can of Coke, every bar of soap, every shirt, every shoe, every CD and DVD, everything you can think of, even pets and cattle -- carried an electronic tag with a unique identifier. But it turns out that a world of electronically identifiable items would be beneficial to more than just machines. Many large retail firms -- including Wal-Mart, Procter & Gamble, and Gillette -- have embraced this vision; they say that it will do nothing less than revolutionize retail sales, and they're spending a lot of money to make the idea a reality.
But even as computer scientists wax lyrical over the prospects of a network where everything is broadcasting to everything else, and companies salivate at the chance to have total inventory control, privacy-concerned consumers are looking with alarm at a technology that promises levels of surveillance unprecedented in human history. It's bad enough, for some people, that a grocery store can assemble a database of your habits from everything that you've rung up on your credit card in the last year. But in a world where every product is embedded with identifying technology, the possibility would exist for companies to know what you're doing, in real time, in almost every aspect of life.
That world is on its way. In 1999, with the participation of several corporations, Brock and Sarma set up the Auto-ID Center, an MIT consortium whose goal is to create an inexpensive, industry-standard product-tagging system using a technology called radio frequency identification, or RFID. By 2006, Wal-Mart plans to use the center's RFID technology to track all shipments moving through its supply chain, the path from factory to store warehouse. Many firms, including the Gap and the U.K. supermarket chain Tesco, have tested systems that can track items as they move within individual stores, alerting employees if products are misplaced or are being stolen. Proponents of the technology say that RFID offers many benefits to consumers as well. Tagged items could enable quite extraordinary new consumer devices; you'd have smart shopping carts to calculate the nutritional value of the food you buy and to automatically check you out of a store, or smart washing machines to alert you if you're washing whites with colors.
But are such benefits worth the price? For the past couple of years, a small band of activists has been mounting a vociferous campaign against radio tags -- they worry that tagged products will show up on store shelves without our knowledge, that we'll be tracked through stores and in the world without our consent, and that, in the worst case, brave-new-world-type scenarios will become an everyday reality. Katherine Albrecht, the head of a group called Consumers Against Supermarket Privacy Invasion and Numbering (CASPIAN) and RFID's loudest critic, often warns of the possibility of RFID getting into the hands of a dictatorial regime. What would a Saddam Hussein do with RFID?
Albrecht is tenacious, and her work has caused some embarrassment for the Auto-ID Center. She recently discovered an article in Smart Labels Analyst magazine, a subscription-only trade publication, that described an alarming RFID setup at a Tesco store in Cambridge, England. According to the Smart Labels Analyst article (which Albrecht read to Salon over the phone), a surveillance camera trained on Gillette razors was activated each time a customer removed a package of tagged razors from an RFID "smart shelf"; the system was apparently taking pictures of each razor-blade buyer (or browser) to prevent theft of the Gillette Mach 3 blades, the world's most-stolen retail product. Albrecht first alerted the (London) Guardian to the camera-enabled-shelf story; the newspaper reported the news on July 19, and within a few hours the story made its way to many blogs and discussion sites like Slashdot, where hundreds of readers railed against RFID.
For Albrecht, the Cambridge Tesco incident highlights the main concern of people fighting RFID -- we won't know where these tagged products are, they say, and we won't know how they're being used.
Proponents of tagging say that the fears are easily assuaged. Kevin Ashton, the executive director of the Auto-ID Center, pledges that RFID technology will be used carefully: You'll be told that your items are tagged, you'll be given the choice to disable the tag when you leave a store, and your name will not be tied to the products you've purchased. If those guidelines are followed, what's so bad about RFID?
RFID, like cloning or genetically modified foods, promises to be one of those technological advances that could remake, for the better, everything about how we live our daily lives. The technology is not just about making shopping cheaper and more pleasant -- although, considering how much time Americans spend shopping, that would be good enough. But because RFID adds intelligence to the things around us, it would usher in an era of networked objects -- an "Internet of things," as the Auto-ID Center says -- that would change the world dramatically: packages that know how to sort themselves for recycling, meat that alerts you when it's been recalled. So far, the debate surrounding its use has also echoed the debate around biotech. Opponents are long on speculation as to all the ways RFID could become disastrous, while proponents may be naive in their expectations that Americans will accept RFID without a fight. But what seems indisputable is the reality that this fully networked world is on its way, and we're going to have to learn how to live in it.
Radio frequency tagging is not exactly new technology. According to Bill Allen, a marketing executive at Texas Instruments, which makes key bits of RFID systems, radio tags were first used in the Second World War. "Each Allied aircraft was fitted with a transponder, which sends and receives signals," Allen says. "This was done to prevent friendly air battles -- if you get a friendly signal from an aircraft, you do not shoot." Conceptually, the tags have not changed a great deal since then; the main difference between the tags of today and the tags of the 1940s is size. Today's tags are tiny -- as big as a dime or as small as a grain of rice, though the size varies with the specific application.
Not only are radio tags not new, but they're also not very strange. Millions of people already use radio-tagged devices every day. RFID is found in electronic toll systems such as FasTrak and EZ Pass, and in gasoline quick-pay systems like Mobile's Speedpass; and millions of cars are equipped with RFID immobilization, which prevents the engine from starting unless an RFID reader senses a radio tag embedded in the key.
Perhaps the largest application of RFID today is in agriculture. Cows all over the world have radio tags embedded in their shoulders or lodged in their second stomachs, and as they move through the various stages of their factory-farm lives, the radio tags let farmers know how they're doing. "You're recording data points," Allen says. "You record the cow's lineage, its shot records, feed lots that it may have visited." This sort of tracking becomes indispensable when problems arise. If you're trying to isolate a sick cow, or trying to determine which of your animals may have feasted on some mad-cow-infected rendered-protein feed, RFID will help.
It was only in the 1990s that businesses began thinking about using radio technology in retail sales. At the time, Kevin Ashton, the Auto-ID Center's director, was a brand management executive at Procter & Gamble. "We started to notice some fundamental problems in our business," he says. "It first manifested itself in products not being on the shelves. I would go to check the stores, and I noticed some of the best products, our beauty products, would be out of stock in four out of 10 stores at any one moment. That's a real moment of clarification. The average industry out-of-stock number is 10 percent, but what we actually found was that the extreme cases tended to be bigger, and that was usually the case for the products that we cared about the most. If the product is really good, that's going to drive it off the shelf."
He adds: "The root of the problem turned out to be information. There wasn't enough information about where things were to make good decisions."
All the apparent orderliness of a typical retail store, with a place for everything and everything in its place, is a bit of an illusion. Once a product gets out into the din of a busy store, it could go anywhere: Customers or employees may move it about, putting it in the wrong section, or they may steal it; or the employees might forget to restock the shelves, or they may make some other error. Keeping things orderly is very costly, and it's slow. In most stores, Bill Allen says, "an inventory is periodically taken where they have to shut down and pay people overtime. It can be very labor intensive and cost inefficient, and even then, the Gap says that any type of physical inventory they do, they guess it's only about 93 percent accurate."
The supply chain, the back-end machine of major retailers, is also prone to errors. Paul Fox, a spokesman for Gillette, one of the main sponsors of the Auto-ID Centers, explains: "Imagine I ship you 100 cases, but when you record it you put it down as 10. So now your inventory system has effectively lost 90 cases." This means that your computer will tell you that you've run out of the product long before you actually have, so you'll order more than you need -- and your supplier will draw erroneous conclusions about the demand for the item in your store, giving a distorted picture of how the product is doing. Companies are loath to operate in this dark, data-less void; they'd do anything for more information, because any data they get could save them millions.
"At any given point in time, the Gap knows where 85 percent of their inventory is," Allen says. "You might think that's pretty good, but then you wonder what that 15 percent represents for a company as large as the Gap. What it multiplies out to is $1.7 billion of inventory that they don't know where it is. It could be stolen within the supply chain and they'd never know it, and on and on and on. What it means is they have to buy more inventory to make sure they have their supply stocked -- and that is a cash-consuming type of situation."
At least for the next couple of years, most companies are looking at this back-end part of their businesses as a main target for RFID improvements. In the retail world, Wal-Mart's supply chain is regarded as by far the most advanced, and its decision to use radio tagging is likely to spur its rivals to do the same. Tom Williams, a company spokesman, says that Wal-Mart is convinced it can see huge savings by switching to this system. "We have 103 distribution centers throughout the U.S.," he says, "and many of them are over a million square feet. When products come into the center we track them -- today, we primarily rely greatly on bar-coding, and as you know, bar code scanning is a step-by-step process. RFID is all at once, and for us that's really crucial for fast-moving merchandise."
The company -- which has asked all its suppliers to put radio tags on pallets and cases they ship to Wal-Mart within the next two years -- will also track RFID-tagged merchandise as it moves on trucks around the country, which will allow it to more efficiently route stock to where it's needed most. In addition, Williams says, the mountains of data the firm gleans from all these tagged cases moving through its operations will "give us a lot of numbers to crunch, and we'll be able to spot trends we don't even know about."
But in Wal-Mart's case, for now, the tracking will stop in store warehouses. Individual products will not be tagged, and although the company has long been rumored to be working on plans to bring RFID into stores, Tom Williams, the spokesman, wouldn't confirm or deny anything. "I think the issue of item-level tagging is quite a ways down the road," he said. "Our immediate effort is with the supply chain."
Williams' reticence may be an indication of how nervous companies are about the prospect of bringing tagged products into stores. Although Bill Allen, at Texas Instruments, went into detail about the Gap's tests of the RFID at some of its stores, a Gap spokeswoman declined to divulge any specifics about its plans for the technology. She confirmed that the company had run some tests, but she would not say where or when they'd been done, or what the results were.
According to Allen, though, the Gap had tremendous success with RFID. "They did a study of four stores that fit the same demographics," he says. "One had RFID and the other three were a control. They looked at sales data in the stores from before, during and after the test, and what they saw was a 15 percent increase in sales in the store that had RFID. They determined that the main reason for the increase was because more of the merchandise was in the right place at the right time."
For retailers, this is the key promise of RFID -- better control of inventory, leading to more efficiency, less waste, and higher profits.
But what will we, the customers, get out of this? Early in July, Katherine Albrecht, of CASPIAN, announced that she'd found several "confidential" documents lying in public view on the Auto-ID Center's Web site. Among these was a survey of consumer attitudes toward RFID that she thinks ought to give proponents of the technology a lot to think about. (The Auto-ID Center says the documents weren't confidential, just mislabeled.)
In focus groups the Auto-ID Center held in the United States, the United Kingdom, France, Germany and Japan, it found that many people saw no obvious benefits to RFID. "If consumers are made aware of any negatives (in the real world this could happen through negative press coverage) they have no benefits to balance their feelings against," wrote Helen Dulce, the Center's European director, in a research presentation. "For example, in Europe there is a large controversy over the health dangers of mobile phones, however mobile phone usage is on the up. This is because this technology has many benefits to consumers (convenience) and these benefits clearly over rule the very strong negatives. In the case of EPC network [the Auto-ID Center's name for its RFID technology] there are currently no clear benefits by which to balance even the mildest negative, so any negative press coverage, no matter how mild, would shift the neutral to a negative." (A PDF of the presentation -- a huge file -- is here.)
But it's not true that we'll get nothing out of RFID. For one thing, even though Albrecht is dubious of the claim, radio tags seem certain to save us money. Wal-Mart, whose raison d'être is low prices, would not remake its back office for RFID if the technology couldn't cut costs. Activists say that even if the companies do save money by using RFID, the firms won't pass along the savings to us and will just keep bigger profits for themselves -- but in the world of retail sales, lower prices are the key to winning market battles, and anything a company can do to cut costs will be reflected at the checkout counter.
The best argument critics of RFID have when presented with the idea that the technology will reduce prices is this one: It's not worth it. "To save 10 cents on a pack of Gillette I'm going to allow them to track me?" Albrecht asks incredulously. "I think many people will say, 'I think today I'll buy Schick.'"
The Auto-ID Center's numbers seem to support Albrecht's claim. In November 2001, Fleishman-Hillard, the center's public relations firm, conducted a small survey to gauge how people would respond to RFID the first time they heard about it. Of 317 people surveyed, more than 80 percent said that they grasped the technology and could see some benefits to it, but 78 percent also had privacy concerns. According to a presentation Albrecht found on the Auto-ID Center's site, half the people said they were very worried about the implications of tagging; 15 people used the phrase "big brother," and telling people that the tags could be shut off was "not compelling." (Here is a PDF of the survey, also a large file.)
Why do people react this way to the idea of their products being tagged, even if they're reassured that the tags will be shut off when they leave a store? One reason could be that people don't believe that the tags will actually be shut off. Kevin Ashton, the Auto-ID Center director, is insistent on that score: "The technology we built into this, called 'kill,' does exactly what it says. On receipt of that command, the tags will self-destruct," he says. "Now, they don't actually blow up, of course -- but the tag will blow a tiny fuse and will be rendered physically incapable of receiving or sending any signals. It's as useful as a light bulb that's been blown."
The trouble is, you'll have to trust retailers about this. If you ask a store to kill the tags on everything you buy, you won't know that they've actually done it. And Albrecht, for one, does not trust that the stores will do what consumers want.
Ashton says that the market will take care of such problems. He recommends that every company using RFID agree to three broad principles -- notification, choice and anonymity, he calls them. Items that are tagged will be stamped with a logo saying so. You'll have the choice to kill the tag when you leave the store. And the tag will never be associated with your name. If you buy a shirt from Wal-Mart using a credit card, for example, the card company will not record the tag of your shirt.
But depending on an industry to police itself is always a dicey proposition. The Gillette test at Tesco offers a case study in why there probably should be laws to police how RFID is used. According to the Guardian and Smart Labels Analyst, the shelf that activated a closed-circuit camera when Gillette razors were removed did not notify customers that it was set up that way. Why not? What happened to Ashton's policy of notification? That's not clear. When called for comment, Paul Fox, of Gillette, said it was not his place to talk about this specific test -- "Those were Tesco's shelves," he said, characterizing the trial as a third-party affair that was completely out of Gillette's hands. The company, he said, had no interest at all in tracking goods within stores. "Our focus in establishing the feasibility of this technology is a very defined window," Fox said. "That window relates to the supply chain, and our only interest rests from the point of manufacture to the retail shelf. We have no interest in the application beyond that." (Tesco did not respond to Salon's requests for comment.)
When told about Gillette's response, Albrecht burst out laughing. She pointed out that Gillette is one of the main sponsors of the Auto-ID Center, whose explicit goal is to one day have most everything tagged. In many of the documents on the Auto-ID Center's site, Gillette razors are held up as an example of a good that could be better managed in stores with RFID. So for Gillette to say, now, that it only cares about the supply chain seems somewhat disingenuous.
When asked about the Tesco test, Ashton seemed to demur. "Ideally," he said, such a shelf should tell people what it's doing. If it was me doing it, I would have had some sort of notice." But he said that in a test situation, a notice was not as important as in a full-scale rollout.
The more thorny ethical question is this one: If the shelf did tell you that it would record your picture if you removed a package of razors, would it then be OK? Albrecht says it would still be terrible. It's much worse than the ambient surveillance system that already pervades our stores, she says, because "this technology is presuming that every single Gillette customer is a criminal."
But don't all surveillance systems presume that you're a criminal? That's why they're watching you, right? Because you might do something naughty. And isn't this system, which focuses on just on heavily stolen goods, actually better, since it's forgetting about all the other people in the store? That's the argument that Ashton makes. The system, he suggests, might be beneficial to all the razor shoppers who aren't stealing, as there would be some record of their innocence; such a shelf would prevent false accusations. And, Ashton notes, if such a shelf helps stop razor thefts, he won't apologize for that. "Stopping people from committing crimes is a good thing."
This is not an argument with which Ashton will change many people's minds, though. There is something viscerally disconcerting about a product triggering a camera to take your picture, however rational it may seem. And even if the tags on all your items have been killed when you go out the shop door, there'll be something disconcerting about knowing that, maybe, someone is tracking your new shoes.
Ashton recognizes that people will feel this way. He says he knows that people have concerns about RFID and that retail firms will face a huge battle in getting people to embrace it. But he has faith that people will come around. He doesn't have any real fears that -- as happened with genetically modified foods, say -- public concern will greatly slow down radio tagging, because he believes his industry will always be upfront about what it's doing. And, he says, "I think consumers are intelligent enough to see how good this will be."