The Internet of Things is Becoming a Nervous System
In 1999, Kevin Ashton (top) was an assistant brand manager at Procter & Gamble when he discovered the final piece of a mental puzzle he’d been constructing over the last few years. Coming up with great products consumers wanted was tough enough, but keeping them stocked on store shelves after they took off was even more vexing.
The problem got Ashton thinking. What if you could track products with sensors connected to the Internet? That way a company would have a real-time picture of its global supply chain, helping it to make sure consumers could
Ashton floated his vision in a presentation titled, “The Internet of Things.” It captured P&G’s imagination. The company loaned out Ashton to the Massachusetts Institute of Technology so he could develop the idea as executive director of MIT’s Auto-ID Center.
As you combine information from multiple sources and they are synchronized on the same timeline, you’ll start to see machines making smarter and smarter decisions, anticipating things to a high degree.
Under Ashton, the Center blossomed into a global organization with laboratories at six major universities and more than 100 corporate sponsors, including Walmart. Now, 15 years later, Ashton’s vision has become the force propelling what many people in Silicon Valley say is the Internet’s third wave, as tens of billions of devices connect to the Internet and create a sort of global nervous system.
As we advance into a new Age of Intelligence™, where machines and devices don’t only communicate but enable human action and interaction, Ashton revisits his singular insight, explains how it has evolved and offers his vision for where intelligent things are headed. This conversation has been edited for length and clarity.
The Birth of the Internet of Things
I joined P&G out of college in brand management and was on a team back in the mid-90s that worked on Oil of Olay. We launched the Oil of Olay color cosmetics branch, which was a lot of work. But after the product launched, I would go to my grocery store, and I didn’t see the product on the shelves or the right colors on the shelves. Initially, I was told this was an exceptional situation, which seemed kind of improbable.
In fact, four out of 10 stores didn't have the right product on the shelf. What I gradually figured out was that stores then used bar codes to try to manage inventory, which gave them a big picture, but it didn't help the guys understand what they needed to put on shelves.
All of the information in computers was coming from people. People are not good at gathering information in the real world. As a result there were huge information holes when it came to real world information. That was the first piece.
So I did a little work with some guys at MIT. We figured out the Internet was the answer.
Sensing is a network technology found in all creatures. You have sensors all over your skin. All of that distributed sensing gets aggregated in your brain. Your brain figures out what’s going on in the world based on these inputs. The Internet as a network had the potential to become like a nervous system if we added sensing capability.
The phrase “Internet of Things” was the title of a PowerPoint presentation I made in early 1999. I’d been pitching this around P&G for a year at that point. The senior guys were finally getting excited about the Internet. “Things” was a word other people in the embedded computing field were already using. There was a professor at MIT, Neil Gershenfeld, who wrote a book called When Things Start to Think. So really my contribution was to join “Internet” with “things.” My real creation was the word “of.” Having made the case at P&G I got this opportunity to lead a research program at MIT to make the IoT real.
IoT: 15 Years Later
There was no Wi-Fi in 1999. There was no email on your cell phone. GPS did not exist in any meaningful way. These technologies were very rare and expensive. Today, there are a couple of billion smartphones sold annually, and 3 billion RFID tags sold last year. There were practically zero in 1999. In 1999 you went to Blockbuster and got a videocassette.
When you think about all that, and where we are today, most people have a device in their pocket that is a very accurate GPS device and is pretty much constantly connected to the Internet. What’s happened in the last 15 years is this massive proliferation of wireless network sensors. It is incomprehensible how much things have changed in the last 15 years.
Cost, Security, and the Promiscuity Problem
The hard problem today is less about the hardware and making low cost sensors. It’s more about processing this unprecedented amount of streaming data in real time and making good choices based on an ever-changing picture of the world. It’s kind of like how the brain works. The data science—that is the big problem at the frontier of computer science today.
The Internet as a network had the potential to become like a nervous system if we added sensing capability.
The problem we have now with security in some of the IoT is that some of these more basic devices like RFID tags will share their info with anybody. There is no authentication. People who are security experts are always concerned when a device is promiscuous. But as we get more computer power for the same price we can start to solve those problems. Security is like an arms race. Do the security and privacy risks outweigh the benefits? Absolutely not. The benefits are many millions of times greater.
The Opportunity for Intelligence
When businesses are thinking about opportunities they are thinking about what product can they make and sell. The biggest business opportunities are not necessarily in new products and services. They are probably in new infrastructure you can invest in and deploy to improve how you sell your other products and services. We have to get much more efficient at how we manufacture and distribute those things. It’s not just an external opportunity. It’s an internal opportunity.
There are a couple of predictions that are easy to make but very difficult for people to believe. We are moving to a world where computing doesn’t need electricity.
People have heard about Moore’s Law. The law that is becoming more important in the IoT world is Koomey’s Law. He’s a guy at Stanford. Koomey’s Law says since the 1950s, the amount of electricity needed to do a computation halves every 18 months. So for fairly simple computational processes you need in wireless sensors, it is gradually becoming possible to have a small battery or in some cases an RFID tag with no battery. Most RFID tags get power from the radio waves talking to them. Batteryless computing and power-less computing is something we’ll see more and more of. This leads us to disposable computing and ridiculously ubiquitous computing.
Another thing that will be very surprising is the rapid pace at which machines are able to make very good decisions. Their ability has been previously limited by the information available to them. As you combine information from multiple sources and they are synchronized on the same timeline, you’ll start to see machines making smarter and smarter decisions, anticipating things to a high degree.
You have to be real careful with the definition of intelligence. To some people it is analogous with the processing power of a device—how fast they did math. But an RFID tag knows its name and when it is asked it can tell you its name. The RFID tag has the behavioral intelligence of a kindergartner. What’s actually become more intelligent is not necessarily the device but the network, the system as a whole. Intelligent behavior results when you have information coming in from multiple parts of the network and being acted on by some processing capability. Intelligence is not concentrated; it’s distributed. The reality is that intelligence is a function of the nervous system.