A reader sent me a link to a video interview with Tim O’Reilly, founder of O’Reilly Media, in which he states: “I think that RFID is an evolutionary dead end, and the reason I think so is the reason why we don’t wear name tags all the time. You think about this conference. People have name tags because they have to be identified so they can get into sessions… but that’s not in everyday life. If we meet again, I won’t recognize you because you are wearing a name tag. I’ll recognize you because I’ve seen you before. Our machines are getting like that. Semantic Web or RFID is things wearing name tags, and Web 2.0 is learning to recognize things by accumulating context… We’re getting to that kind of augmented reality, where our computers will have senses that are as good as ours or better… they are going to recognize faces, they are going to recognize objects, they are going to have immediate recall.” (See Tim O’Reilly on Recognition, RFID and Web 3.0.)
I don’t know O’Reilly personally. I know he’s very successful and well-respected, but I think, frankly, he needs to get out from behind his Web 2.0 computer and step into the real world. He also needs to think through his opinions a little more deeply. His own argument against RFID proves the case for it.
O’Reilly claims we don’t wear name tags in everyday life, and that’s why we don’t need RFID. In reality, however, we do wear name tags every time we are faced with an unfamiliar situation in life—when, for instance, we visit our kids’ teachers at school, go to a conference, or enter a building where we don’t work. We also drive around every day in cars containing license plates that allow us to be identified.
The reason is that as good as human senses are, they can’t identify what they don’t know. One of the big things RFID does is enable computers to identify things they’ve never seen before—so, for example, a customs agent can identify a new product you’re sending to a new customer.
Another problem with O’Reilly’s argument is that as good as human senses are, they can’t easily distinguish between different objects that look identical. One reason food companies are turning to RFID is that the technology can quickly tell workers which of three identical containers holds the sugar or flour that’s been sitting in storage the longest. And apparel companies are turning to RFID because it can help a salesperson identify a medium-size shirt that has been placed on a shelf with large shirts.
It’s hard to believe computers are going to get so sophisticated that they will be able to point a camera at a shelf of red polo shirts and distinguish the medium-size shirts from the large ones—or know which corrugate box of soup goes to Wal-Mart and which to Albertson’s, or which plastic container of sugar is the oldest and should, therefore, be used first.
Another issue has to do with the limits of some types of senses. We can view and recognize objects with great sophistication, but we can’t see into boxes, and neither can a computer equipped with a camera. An RFID system, however, can identify items within a box, which happens to be an extremely useful ability in the supply chain.
RFID systems are sensory systems that enable computers to monitor products, assets or the environment on a large scale. A single computer, for instance, could be connected to an RFID network and monitor hundreds of shelves in a warehouse or store. That seems far more cost-effective than deploying computers everywhere to monitor the shelves with cameras and other sensors.
I do agree with O’Reilly’s view that we are moving away from a world in which people look at reports from computers and make decisions, to one in which computers make decisions and people continually improve the algorithms the machines use to make better decisions. But I think what O’Reilly is missing is that RFID is one very critical type of sensor for a computer, and that a computer’s sensors will not necessarily be modeled on our own. We can recognize people through facial recognition, and so can a computer. But that doesn’t mean a computer can’t recognize objects by reading a serial number on an RFID tag. Computers will naturally have sensors we don’t, and those sensors can be networked to cover large areas cost-effectively.
Call it Computer 2.0.
Mark Roberti is the founder and editor of RFID Journal. If you would like to comment on this article, click on the link below. To read more of Mark’s opinions, visit the RFID Journal Blog or click here.