38 min . Oct 19, 2026 . Engineering Materials Science and Engineering
From the cards we tap at checkout to the cars and appliances we control from our phones, everyday objects are increasingly connected. We speak with Sanjay Sarma, a professor of mechanical engineering at MIT, about how these objects communicate and what it takes to get them speaking the same language. Sanjay helped develop the RFID standards that allow retailers to identify and track individual products. He shares how keeping store shelves stocked helped drive the technology, what digital twins let us monitor and control, and why bringing artificial intelligence into the physical world raises new questions about safety and reliability.
More From Sanjay Sarma: https://learn.mit.edu/search?resource=3437&resource_title=special-topics-in-supply-chain-management
Host: John Harrold, PhD
Opening excerpt:
Sanjay Sarma: Just keeping track of them is so chaotic. Retail is like a war zone. And that's where it took off.
Conversation:
John Harrold: Hello, I'm John Harrold, and this is Science of Everyday Life, where we get to explore some of the science underlying the things that make everyday life possible. Today, I'm here with Sanjay Sarma, a professor of mechanical engineering, to talk about how some of the objects in our lives talk to each other. Sanjay, welcome.
Sanjay Sarma: Thank you, John. Pleasure to be here.
John Harrold: Great. So I think the first thing we should do is kind of set the stage. Can you tell us what the terms RFID, NFC, and IoT are? Because you get this jargon. But I want to, I want to hear it from the source.
Sanjay Sarma: Yeah, yeah. You know, RFID is when you drive through it, you know, on the highway, and there's a toll pass. That's RFID. RFID is used in retail. You know, if you go to a Uniqlo and you do automatic checkout, that's RFID. And a lot of RFID is used in retail and in toll passes and other applications. NFC is a cousin. It's not exactly the same thing. RFID, you can read the tag from a distance. But your credit card, you don't want someone reading it from a distance.
John Harrold: No. That'd be a problem.
Sanjay Sarma: So NFC is near-field. It only works very close. So you have to tap it on your, you know, when you buy coffee, you tap it on the, on the cash,the payment system. And that's NFC, near-field communications. RFID is radio frequency identification. They're all radio frequency, but one is a wavelength that goes long and one is a, is a technology that works in a short, a short distance.
But those are only two protocols. Two communication protocols. You also have Wi-Fi. We have Bluetooth. So my Apple Watch connects to my iPhone over Bluetooth. Right?
John Harrold: Yeah.
Sanjay Sarma: And over Wi-Fi. So it sort of picks and chooses and all these. And of course we have the internet. So all these techniques, all these technologies enable a really amazing way for us to control our environment. So in my home, a lock is IoT, so I can unlock the door. You know, I'm taking a delivery in the garage if I'm taking a delivery or if we have a worker come to the house, I can unlock my door remotely. We have cameras. We have smoke sensors and IoT. So that's Internet of Things. The idea of connected things, keeping an eye on stuff for you and keeping you safe and giving you convenience. I can unlock my car remotely if I have to. I can lock it. If I forget to lock it, I can lock it remotely. I can see where it is. This is all IoT, right?
And then my car. When my car approaches the garage, the garage door opens automatically. That's very convenient. That's IoT. So these are all examples of the Internet of Things which are based on these communication protocols.
John Harrold: And so a little bit about how this works. So power is really important, right? Because your watch doesn't have a big battery in it. And some of these things sending signals sometimes takes a lot of power. And I know it's important. So how do these devices work in such a way where they are able to work on such small scales with such low power?
Sanjay Sarma: There's always a trade-off in engineering as a whole, right? It's power. It's range, and it's bandwidth. That's how much data you can pump through. So RFID tags like toll passes, the tag has no power. The tag actually picks up, scavenges power from the reader. So the reader sends out a signal. That signal is both read by the tag and it powers the tag. And then the tag communicates back with the reader. Actually, it's like a mirror. You know, turning a mirror on and off, so the tag has zero power. And therefore, because it has no battery, it lasts forever.
That's RFID. NFC is similar. NFC. So your credit card has no power. Now your phone may have power, but the credit card doesn't. But NFC, now, in RFID, all it's getting in is an ID. In NFC, it's a little bit more secure. It's your cash, right? It's your credit card. So there's a, there's some security that happens. And there's a little bit more computing that happens, has to happen on the card to make sure that the verification occurs, right? By the way, the same NFC is used by smartphones as well, which do have power, but it's actually broadly designed for these contactless cards. Right. And I mean, these are near-field protocols. It's the way to think about it. But there's a little bit more communication, there's a little bit more compute. And because it's very close to the reader, the, it can draw power and do the computation.
Bluetooth generally, there are now Bluetooth systems that can scavenge power, but generally they're attached to things that have batteries. So my Apple Watch has a battery and it has Bluetooth. So even if I turn Wi-Fi off it can communicate with my phone using Bluetooth. You know, your headphones, those are Bluetooth. They do more bandwidth, but they require batteries and therefore they have to be charged. You don't charge your credit card, but you charge your headphones, right? That's Bluetooth. It's a personal area network, essentially, or a body area network.
And then you have Wi-Fi, which requires much more power. And by the way, the you have cellular, which is a whole other radio frequency technology, but that's yeah, long range. Right? Wi-Fi is much higher bandwidth than cellular. But now with 5G, cell has pretty good bandwidth. So that's but that's like a kilometer or half a mile or a mile.
John Harrold: And so when we're dealing with all this, these are all basically light, right? This is the same light frequencies. Like, are these all radio waves? Are they different types of transmission?
Sanjay Sarma: Yeah. So light is an example of an electromagnetic wave.
John Harrold: Electromagnetic?
Sanjay Sarma: Yeah.
John Harrold: I guess it's more specific.
Sanjay Sarma: It's all electromagnetic. When the frequency is high enough, it becomes light, then it becomes even higher, becomes X-ray. But when it's less than light, it's infrared. Right. So infrared is on the red end of the spectrum. There's near infrared, there's far infrared. But then eventually you get a radio waves and radio waves. You have things like microwave and you have, high frequency, ultra high frequency UHF, for those of you, for those of, for the listeners a little bit older might remember UHF frequencies and so on VHF. So that's that whole spectrum. It's called radio waves. RF is just that when the frequency becomes high enough, it becomes infrared and then light.
John Harrold: And then light. Yeah.
Sanjay Sarma: And, and then X-ray.
John Harrold: Okay. And so and it's, it's cool how we can just tune these things to be used for, for different applications. And you know, you know, the spectrum.
Sanjay Sarma: You use a different part of the spectrum depending on what you're trying to do. Because different at different frequencies, you get different propagation properties. Very long radio waves can go around walls and, you know, that penetrate the higher the frequency, the more it behaves like light. You have shadows. It doesn't get through stuff. Right. And these these radio waves, the frequency at which you operate, as I said, it affects the spectrum, it affects your penetration propagation, but also it's highly regulated by the government. So some parts of the spectrum are given to the military. Some parts are given. They used to be given to radio right before digital radio. Before digital. I'm sorry. TV, they had a lot of spectrum. Radio has spectrum.
And that's like highways. That's licensed. Private companies license it. So cell phone companies will license it. Some of it is meant for industrial, scientific and medical. That's called ISM. That's more free. We all can use it. And that's where, RFID tends to operate and Wi-Fi for that matter, and Bluetooth.
John Harrold: And so we're talking about these regulations in the space. And... How did you get involved in this? I know this technology predates, you know, a lot of your career. Right. This is pretty old. I think the first use I was looking it up first commercial use of RFID, is was in a Star Wars toy. Right? At least that's what Wikipedia said. I hope it's right. In, like, 1980s, early '80s. But you came in and... Well, how did you, you come in, start standardizing these things or working with these things?
Sanjay Sarma: RFID is very old. I mean, radio frequency, actually, MIT has a long history. Lincoln Labs is there for radar, right? So radar is basically you send radio waves out and if there's an object that reflects it, you get a signal back. And that's how you detect that something's out there. And so the idea of putting transponders on friendly aircraft so you could distinguish them from any enemy aircraft, it's called, identify friend or foe, IFF. That's actually an example of an RFID tag that aircraft have, and that's, that's 70 years old.
John Harrold: Oh, wow.
Sanjay Sarma: Passive RFID happened more recently because low-power circuitry became possible. And that showed up in a bunch of applications: toys, bus tickets, etc. And then I got involved because actually a colleague of mine at MIT, a gentleman by the name of David Brock, was like, "Sanjay, why don't you put this on everything?" Right? No, I can't do it. You know, it's difficult. And also at that time and there were lots of different ways to do it. There are lots of different standards. The performance was poor.
But, once we got into it, we realized, I realized that you could rethink the protocol completely, the RFID protocol, and you could rethink, the way the chips are designed to make them cheap and low power. And then it could. And then David, at that time, was like, "Why don't we use the internet?" Right? So then we thought, how do we use the internet? So I thought, we could have sort of data routers. This is before the cloud. The cloud didn't exist. We used to call it the sky.
John Harrold: Yeah, right.
Sanjay Sarma: So we decided that we could create these data routers in the sky. So in many ways, we anticipated, the cloud, we called it the sky. And so we put data up there but not on the tag. And then we came up with this concept of the Electronic Product Code, which is just a license plate on the tag. David gets a lot of credit. And I think that between David and me, there was another man, Sunny Siu, and another, dear colleague, Kevin Ashton. He came up with the term IoT because once we were thinking about all these things, communicating with each other, then he said, "Why don't, " he came up with the concept of calling it the Internet of Things, the idea that everything would be automated and connected to each other. So it's a group effort.
And then there, we started a, a research center with hundreds of companies involved. And there were some amazing engineers from these companies who got involved, you know, who all brought innovations and different ideas. Steve Smith, who's now at Apple. Chris Diorio, who's the CEO and founder of a company called Impinj, which is now public. Just amazing people, you know.
John Harrold: So when there's a new technology or something happening, in order for people to utilize it, there has to be a reason to change. It's got to be something that isn't working right or something that people really want that isn't possible. So in those early days, what was the thing that really drove this to adoption? What was the problem that you were really trying to solve?
Sanjay Sarma: I'll give you, a sort of a romantic version of it. It was much more complicated. But the application that we went after was retail. So retail has several issues. The problem everyone talks about is checkout. But that's really not the application that drove RFID. But that's a very evocative thing. You just leave the store with stuff and it knows what you took. You don't have to wait in line. So that's the, the flagship, you know, use case.
But the ones that were really concerning to retail one was theft. Right. And the other was actually even more, even less glamorous, but even more important, which is just keeping track of inventory. So retail has all sorts of problems. I mean, this colleague of ours, Kevin Ashton, used to be at Procter & Gamble, and he was one of the people who talked about the fact that, you if you go to a store and you don't find Colgate toothbrushes, right, you're going to buy a competitor's toothbrush. So Colgate loses out and often you'll go to a different store. If you care about Colgate, you're going to go to a different store. The retailer loses out. So just making sure that inventory is on the shelf is very painful. And you got to have all these people look for inventory and make sure that the shelves are full.
And, you know, it's called shelf replenishment. So that is the application. And this becomes especially bad with clothing, because if you have a bunch of shirts small, medium, large, how do you know the larges are there or the 42 regular is there? You know, just keeping track of them is so chaotic. Retail is like a war zone. And that's where it took off. Then it took off and things like toll passes.
John Harrold: Oh, interesting.
Sanjay Sarma: But it's the same standard. What we did do was develop a single standard that could support all these applications, make it sort of minimalistic, put data on the cloud. So we made the chips very small. We made them very cheap as a result, you know, a few cents, the tag became a few cents. So it was, you have to do a lot of things to make this work.
Now, I did say I overromanticize. Just the fact is, this was eight years of just, a lot of block and tackling, a lot of friction. Companies that had intellectual property the old way didn't want to give it up. There's a lot of opposition to the standard. There was a lot of intellectual property issues. There's a lot of entrenched interests. You know, it's quite a long fight.
John Harrold: And so without getting too much in the weeds, what is that standard like? Is it like, a language you guys decided it all on, or is it. How would you describe this, what the standard is?
Sanjay Sarma: So, a protocol, captures the protocol, and most, protocols have a number of levels. You know, you need a number. The tag is a number, a license plate, Bluetooth and Wi-Fi. They all have numbers. You know, you have your, something called a MAC number, right? Then you have an SSID, you know, in Wi-Fi, right? So you need sort of those, a nomenclature, naming and numbering things.
Then you have the physical layer, which is which frequency you can operate at, you know, waveforms look like. How do you modulate the waveform? Then you have the logical layer, the commands, you know, which is when you have lots of tags, you got to figure out if they all respond at the same time. It's, it's bedlam. You want the tags to speak to you one at a time. So you got to sort of control the access and say, "Okay, now you may speak, now you may speak, and now you may speak." So all those things are laid out, and you have to be very creative in these protocols to come up with, algorithms that can let you read thousands of tags in a, in a given from a single reader.
John Harrold: And I assume that the level of complexity in the protocol would be, proportional to the cost of the tags fully realized. So if you want to have a lot of information in the protocol, maybe a more expensive tag that maybe will be harder to get.
Sanjay Sarma: That's a question. That's a great question. Let me put it this way. To make very inexpensive tags, the protocol shouldn't be complicated, but needs to be very well thought out and very intelligent. So it's very elegant. It's an elegant protocol. And all these people I named earlier, they all contributed to a very elegant technical set of technical solutions, you know, and we had to think about things like privacy, data security in even the protocol.
Like if I'm at a work in a warehouse and I'm reading all the tags and someone can hear what I'm reading, then I'm doing industrial espionage. So we had to even think about things like that as a community. So the protocol is, for RFID, because you can't, you don't have power, you can't put a lot of computation on the tag, is a very elegant protocol. When you have power, you can be, frankly, a little bit more, expensive. You can make the protocol, you can cut corners, you're not pushed to the limit. But when you have constraints, you have to be very creative with the protocol.
John Harrold: And I guess you can't update it.
Sanjay Sarma: It's hardware.
John Harrold: You set it, right?
Sanjay Sarma: You're stuck with it, right? So you got to get it right. And then there's intellectual property. You got to make sure that as much as possible, although it's impossible to do it completely that the folks who make your tags, right, they aren't violating someone else's patent and getting sued. And, you know, that happens. And so it's quite, it's, it's it's like you have to walk on eggshells and do ballet at the same time to design these things. So I call it ballet on eggshells.
John Harrold: It sounds quite difficult. So this is something that's that happened over 20 years ago, right? 27 years ago, I guess, when you started this, organization. Right. Was it the Auto-ID Center?
Sanjay Sarma: We started it, believe it or not, it was inaugurated in 1999. Well, 27 years ago. Yes.
John Harrold: And this is something that is currently used today. I was reading news articles from 2025 talking about the expansion of RFID in retail. You know? Yeah, but it's already been used for a long time. What's happening right now with RFID? Are they still developing new things? Is interesting stuff happening or is it a technology that's fully realized now? It's infrastructure?
Sanjay Sarma: So, the thing about RFID is that, so an RFID compared to, say, Bluetooth, the readers are fixed and the tags move, you'd have handheld readers, but and readers are more expensive and the tags are ultra cheap, right. So the RFID reader is more expensive than your cell phone or in that range, maybe, maybe cheaper than your cell phone. But it's like that. But all it does is read tags. Basically, right? The tags are extremely cheap, a few cents.
What that means is the infrastructure is expensive. If you wire up a store, the infrastructure is really expensive. So change is not that easy, right? And so, no, there haven't been many changes. There have been new versions of what we call the Gen 2. We had Gen 1 and Gen 2, second-generation RFID. So we're still in Gen 2, a quarter century later, there have been, new versions v1, v2, v3, and these versions are backward compatible. So the tags change. The readers can still, read them and you can update the protocol on the reader, but it's got to read a mixed bag of tags, so you can't change that much. Right.
What has happened in the last few years is our signal processing technology has improved. AI, artificial intelligence, which means that in principle, now we should be looking at a new protocol, the next generation. But there's a lot of inertia because of the fixed infrastructure. So in my lab, actually, we're looking at some of the aspects of the new protocol, what they might be. And there's some very interesting things you can do. But change will take time and pain.
John Harrold: So I keep hearing this phrase digital twin. Right. What is that, and how do the technologies make that possible?
Sanjay Sarma: So, the concept of the digital twin, in some ways precedes what we did. But really, became very important after IoT took off. But the basic idea is that if you're, the U.S. Air Force and you acquire, a new, very expensive, fighter jet from a vendor, right? You need to have a digital twin. You need to know, how it operates. And you need to monitor these things, right? Or a ship, for that matter.
So it sort of goes back to NASA and places like that. You need to have a, you know, this, the space shuttle might be up in space, but you need to have something that replicates what's happening there. So, but with RFID, Bluetooth, Wi-Fi and the internet and the cloud, it became possible to make digital twins of pretty much everything, a digital twin is… So if I take my car, I have a Tesla. I have a, I can see what's happening with my car on my phone because my because a replica of the Tesla is in the cloud. My Tesla. So I know where it is. I know if I left a door open, I know what the battery state of the battery is. I know if there's some problems right now, I have a little bit of a problem with, my seatbelt. Right.
So I know that I need to get the seatbelt, the restraint system, changed, without going to my car. When I was traveling, I was in in Asia. I could see what was going on in my car. Right. In fact, for I was showing it to someone, I logged in and my wife was driving it, right, and I knew it was being driven. Right. So that is a digital twin, but now you can expand it. So, I have an Apple Watch. And so I have, in principle, I could have a replica of myself in the cloud. I have an iPhone. I can see where it is with Find My. I have myself replicated in the cloud. I have an Oura Ring, right? So, my home. Right. I can see if the door is locked. I can see what the temperature in the room is. I can see whether the fridge open.
I can see if the dishwasher or the, sorry, the dryer has finished its run, right? So the idea of a digital twin, it's sort of it to do IoT, it makes sense to replicate stuff in the cloud. And that's what a digital twin is. Now, people have expanded the definition to mean an avatar. Right. I don't completely buy that. But some people talk about the fact that I can make a replica of you that behaves like you. Looks like you, has your mannerisms, sounds like you. And you're here, and there's a version of you, an avatar, that someone's interacting with. Actually, I created one of myself when I was in Malaysia, and it looks like me. It has my gestures. So some people call that a digital twin as well. Some people call a scan of your home. If you put it up for sale, a 3D scan of your home, they call that a digital twin.
I think that's less what I have in mind, but to me, what a digital twin is, is a model of the real thing that is both updating in real time and on which you can do simulations to see what-if scenarios.
John Harrold: And, and to, like, interact with this technology. I was learning about manufacturing and how it can streamline all these things that these big, the big machines that make all the stuff and keep all this really important uptime. They can see, predict problems happening.
Sanjay Sarma: Yeah, that's like my Tesla. I mean, the thing about manufacturing is, it's it sounds more distant. That's why I use the Tesla example so I can unlock my Tesla door. So not only can I monitor my Tesla, I can unlock the door. So if, my kid wants to pick up something, sometimes I'll leave it in my car and say, "Just tell me when you're nearby." She texts me. I unlock the trunk, she takes it from there. Right.
So now imagine doing that with machines, a machine tool, so I can create a digital twin of the whole factory. Right. Imagine during Covid, you know, you could see how your factory was going, doing. Right. If you didn't have to go and you didn't have to go in. So the idea of a digital twin is to do the, make the, create the replica and then the digital twin can also control the real operation. Right.
So for example, let's say the power prices, power pricing is becoming more and more dynamic.
John Harrold: Yeah.
Sanjay Sarma: So you can say, listen, you know, some of these operations like I'll turn the oven off, I can do that in the morning. Right. So a digital twin can talk to a twin of the grid and figure out, negotiate, right? You can put some agentic stuff on top and negotiate when it's going to turn the, the oven on when the prices are low. You can do that with the Tesla. Again, you can say charge it late in the night when the grid is not as congested.
John Harrold: I'm just going to jump in to something that you just said. You talked about agentic. So this technology we've started from, you know, you realize we need to start making standards 27 years ago, which was far away from AI. Right. This was, now AI has become and machine learning has become part of everyone's lives on a daily basis. And there's a lot more people dreaming about what it can do. How has that changed the Internet of Things? Is it is it making the process better, the promise better, or is it changing what we expected from it?
Sanjay Sarma: It's making it more worrisome. So let me just clarify that when we talk about agents, what's an agent? An agent is something that does something for you, right? So I can say to my car, "Charge when the price is low, electric prices are the lowest," right? That's entirely deterministic. I'm saying, "When it's in that slab, charge in that slab," right? That's deterministic. AI is fundamentally non-deterministic. It can hallucinate, right? It can, it... You know, every time you ask ChatGPT to write a poem, it's going to write a different poem. Doesn't write the same poem, right? Write a poem about my cat. It'll be a different poem, right? Because it's almost by definition a little bit probabilistic.
Problem with automation systems is when you introduce probabilistic systems, you have to be very careful because you get unpredictable outcomes and predictability, when you come to hard automation, heavy industry, factories, etc., it's very important, you know. Now, can AI be used? Of course, but in a different way. So, for example, maybe my, my Ring doorbell, There's a lot of controversy about these doorbells now. Right. Has a face recognition system. Okay. That's AI. That's better, as long as it's reliable, right? But do I want an AI system, controlling, I don't know, the operation of a factory? What if it hallucinates? So I got to be careful, right?
The thing about these cyber-physical systems where computers and physical systems come together is, in cyber systems, you worry, here's the other problem with cyber systems: We worry about cybersecurity. In physical systems, we worry about safety. When cybersecurity and safety come together, it's pretty scary. So you've probably seen, the, you know, we're in a, there's a conflict going on, and you've probably seen that water systems and other things have been hacked around the country, right? That's bad enough. Now, if you introduce AI and they make a mistake or they go rogue, what happens? So I worry a little bit about the, interaction of AI with nondeterministic. That's AI. You know, systems with these, IoT and other systems.
If you've ever read 42 by Douglas Adams. Sorry, Hitchhiker's Guide to the Galaxy. 42, Douglas Adams, 42 being, of course, the totemic concept there, in the Hitchhiker's Guide. One of the books, there's an elevator whose name is Marvin, and it has AI, and he tries to talk you out of going to the floor you want to go to. And it's pretty funny.
And that's an example of how AI and physical systems, how they interact, could lead to emergent behaviors we might or might not be happy with.
John Harrold: Where do you see this leading? What's the future of IoT? And what is this? I guess the promise of IoT is this future. What does it look fully realized? Because we have the snapshot that you gave us of your life that's, you know, early adopter of these things. And you know what they're doing. So. So, yeah. Tell me about the future.
Sanjay Sarma: I am cautiously optimistic, but I think society also needs to wake up to the issues. When you when we put all our eggs in the digital basket, things like cybersecurity become extremely scary. When you introduce AI, which is unknown as sometimes in some ways it's, you know, unpredictable things become scary. All that said, I think we're very poor stewards and custodians of our resources. How many lights do we leave on? How many buildings in Cambridge and Boston have the AC cranked up when no one's there? How many rooms are being cooled? You know, when no one, no one's there, right. IoT can really make us much more efficient in everything we do.
And, so I think done right, and it's just convenience. You know, I can literally tell my phone, "Open the garage door." Open the garage door. And I'm leaving every morning. I just say, "Open the garage door." It just opens it. I don't press a button. And next step it can, it should be able to, and I'm working on it, read my calendar and say, "I've got to drive now, so open the garage door," right?
So I think we could do a lot with this. I think the opportunity's there, and we can be much more efficient. Life could be much more seamless. Life has become very complicated, but it could become much more seamless if we make this work. Huge applications for health, especially with the aging populations. Right? I mean, it was only during COVID that we took people's personal pulse oximeters seriously, you know, but now, of course, wearables everywhere. I think there a huge opportunity. So I'm bullish. But I hope society wakes up to the downsides.
John Harrold: And what are the challenges that are genuinely hard? Right. There's getting people to work together and that's that's kind of hard. But what are some of the discoveries that we're we need to make or the edges that we need to overcome?
Sanjay Sarma: I think the most important problems are societal policy frameworks, regulatory, the thinking through the, guardrails. I think those are actually the most difficult ones. And then agreeing to standards. Right. Agreeing to, I'm not suggesting that we will regulate everything sort of without thinking it through. But at the same time, societally, we've got to wake up to what this means. So, so that's, I think, a very important thing.
Now, from a technical perspective, there's lots of exciting stuff happening, you know, more the more the number of devices and sensors and things like that can operate without power, the more we can know. I mean, imagine a sensor in your home that tells you if there's a water leak, you know, a stitch in time saves nine, right? And if you could turn the water mains off because there's a leak until you go home, right?
Imagine and we're working on sensors that can in principle detect, a forest fire early. And just imagine the implications of that. I mean, given that the forest fires right now in Europe, in Canada. Right. So, so I believe that new sensors, new communication technologies with drones, for example. The thing about RFID tags, you can only read them, read them from tens of feet. But imagine a drone with a reader that scans the forest floor, and you have RFID-like sensors that can detect the early embers of a fire. Right.
John Harrold: And how fast could this sensor read it so they have a drone going by? How quickly are they able to go by or area to cover and be able to successfully read the tags? In all these things?
Sanjay Sarma: It's not a matter of how fast you can read it, but can you match the rate at which a drone can cover the area? You want to match the two. Even if this reads it every second and a drone can only fly overhead once in a day. There's no point. Right. So we do have to think that through and create systems. But then we have to implement a system where these drones can be deployed and, you know, and create a market for the sensors to be made inexpensively. The problem is, if you make onesies and twosies, and it remains in research, then, the cost becomes an issue. So you got to think systemically to get the performance and cost to where it needs to be to prevent the fire from spreading. So can I read every second? Yeah, I can.
How much will it cost? Well, how many do you want? Depends on the.. right?
John Harrold: Yeah.
Sanjay Sarma: And, and what's the point of my reading every second if the drone can only fly once a day or once in a week? Right. So you've got to match all that. So it's a systemic solution. And that's what we need to work on.
John Harrold: If you’ll indulge me.
Sanjay Sarma: I’ll always indulge you.
John Harrold: I have a question. I always have a question about policy. So policies and government policies go to the borders. Right. Things that we say in the US, Canada can decide to follow or not. So how does I guess policy and leadership help in this space? Like, you know, you these things travel around the world. People travel around the world. Their objects travel around the world. But how would policy shape that?
Sanjay Sarma: The US undervalues its influence on the world. And actually we are squandering that influence. California has a big influence on the US, California Air Resources Board or Bureau, CARB standards. You know, cars are made to CARB standards, you know. So that's point number one. Point number two, you know, the big centers of regulations are the big population centers. It’s the US, the EU. The EU does its own thing now, China and India. Right. So those are the big and they tend to... Engineers, We speak the same language. Right.
So if it's informed by engineering intuition and engineering analysis, standards tend to converge. Where they diverge is when social or political criteria go the other way. You know, if a totalitarian regime is going to operate differently from, a regime that's very open and the ones in between are all going to be different because their definitions of transparency or their requirements are different. Those are the things that make to make things diverge.
But if the US comes up with a standard, I would say much of the world would take it seriously.
John Harrold: Thank you so much. We're coming to time. But when I end these, I would like to ask, a couple of standard questions. You know, just pull from some standard questions across the board.
Sanjay Sarma: What's my favorite color? Blue.
John Harrold: Well, I'll say my favorite trees are ficus. Oh. Thank you. I've learned a lot from you. I've read your books and we've talked, and I consider you someone who I've learned from and been a mentor for me. So I'd love to know what has been some of the things that your teachers or mentors have told to you that still stick with you? And, and they're part of your everyday life?
Sanjay Sarma: I'll say it in a way that's a little bit irreverent. I think the most important thing for me is curiosity. There's an ad. It's come back, actually, the Dos Equis beer commercial. Right. And the Most Interesting Man in the World. And in the end, he says, "Stay curious, my friend." I love that. I love that. I love that. "Stay curious, my friend." There's a lot of research that shows that curiosity drives dopamine, right? Curiosity keeps us young. It protects us from, you know, some of the neurodegenerative conditions that, become, a big problem recently, somewhat, and we, especially in this new world, we have to keep learning and keep stay on our feet, keep our feet moving.
You know, so I would say that I learned that from my mentors. I continue to learn that. If you ever see me not be, just, you know, remind me. And to me, that is the lifeblood, of, of success and innovation. The other thing I'll say is that, in my view, learning being a continuous learning mode, that is the oxygen of the innovation economy. So curiosity and learning, they go hand in hand.
John Harrold: Well, thank you so much. And if you want to learn more, go to learn.mit.edu. Thank you so much.
Sanjay Sarma: Thanks, John. This is a pleasure.
