The Nostalgic Nerds Podcast
The Nostalgic Nerds Podcast, where we take a deep dive into geek culture, tech evolution, and the impact of the past on today’s digital world.
The Nostalgic Nerds Podcast
S2E13 - Warm Coke and the Internet of Things
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
Does your thermostat know when you're approaching your own front door? Does your watch know you're stressed before you do? When your car rewrites its own software at 3 a.m., do you know what changed?
In 1982, a group of Carnegie Mellon grad students wired a Coke machine to ARPANET because they were tired of walking down the hall to find warm soda. Two questions. Is there Coke? Is it cold? That was the entire revolution.
Marc and Renee trace the line from that hallway to the world we live in now. Mark Weiser's dream of calm, invisible computing at Xerox PARC. RFID tags giving products identities they never asked for. The cloud removing every reason not to collect data. The moment your thermostat stopped being an appliance and became a temperature node in a global behavioral dataset.
Along the way, the Internet of Things went from reporting to deciding. Traffic grids reroute themselves. Buildings adjust before you walk in. Sensors feed models. Models trigger actions. Actions reshape your environment. And somewhere between convenience and autonomy, something changed. It used to be "is the soda cold?" Now it's "who chose the objective function your house is optimizing for, and what does it know about you that you haven't figured out yet?"
Notes - For android users that want to detect smart glasses nearby - https://play.google.com/store/apps/details?id=ch.pocketpc.nearbyglasses&hl=en_GB&pli=1
For Apple users - https://apps.apple.com/us/app/nearby-glasses-original/id6761056896
We'd love to hear from you. Click here to give us ideas on new episodes.
Join Renee and Marc as they discuss tech topics with a view on their nostalgic pasts in tech that help them understand today's challenges and tomorrow's potential.
email us at nostalgicnerdspodcast@gmail.com
Come visit us at https://www.nostalgicnerdspodcast.com/episodes or wherever you get your podcasts.
I still maintain, Marc, that my refrigerator does not need firmware updates. I don't need a smart fridge.
Marc:Or a smart oven or a smart washer.
Renee:Yeah, I don't need—why? Why? Just why?
Marc:I think you're thinking, you know, that it's not really, you know, you're thinking because it's an appliance.
Renee:It is an appliance. It is. It's a dishwasher. It doesn't need to be a wireless dishwasher.
Marc:I don't— it's renee renee don't think of your refrigerator as a refrigerator it's a data collection mode for dairy storage capacity no
Renee:No i refuse to believe my butter tray is part of a distributed system no no.
Marc:Your butter tray absolutely has telemetry potential can you can you imagine Like, like a Samsung, you know, you've got the glass, you know, the glassy in the refrigerator and the the weight on the on the butter tray drops just, you know, just enough.
Renee:It's like, yeah, you need butter, Renee.
Marc:You need. Yeah.
Renee:I hate you, refrigerator. Like, I just don't need one more thing. OK, OK, I will say this. So my washing machine does have wireless capabilities. It can text you when it's done. Okay, when I...
Marc:Okay, okay, that. Yeah. Because the time...
Renee:It's far enough away now. That's right. I don't hear it.
Marc:Well, the amount of time that it says on the little thing, you know, like it says 38 minutes or whatever. Yeah. And you set a timer for 38 minutes and then, you know, came back, it would say, you know, 12 minutes left or something like that. Yeah. Because washing machine time is not the same as real time.
Renee:Oh, there's no atomic clock in there.
Marc:No. Yeah.
Renee:That's fair. So, okay, we agree. The washing machine can email me, but everything else I don't want to hear from. Okay, so let's go back to 1982. Carnegie Mellon University, where all good things happen. It's where all good things start, I think. Okay. Beige walls, fluorescent lighting, academic anxiety in the air, right? There's a Coke machine in the hallway. That Coke machine is the birthplace of the Internet of Things. This is not a TED Talk moment. No one said, we will digitize the physical world. No, no, this was a grad student, and they were tired of walking down the hall only to discover that the soda was either gone, that the soda machine was empty, or all the soda in it was warm, and they would be really, really mad, right? So they wired the machine, and they hooked it up to the ARPANET, which at the time was the internet for a college, right? It was completely closed, but the students had access to it. So they hooked up the ARPANET, and it reported only two things. Is there soda in that machine? And is it cold? That's it. That's it. That's where the whole revolution started from. I will never look at a vending machine the same way again. It was the birthplace of the Internet of Things.
Marc:I'm with you. I've spent so much time on IoT and those simple capabilities, right? So think about these things, right? You've got a physical object exposing internal state over a network protocol. Like that has not happened before. Temperature sensors wired to some sort of logic. I bet you it looked like a rat's nest, you know?
Renee:I can keep thinking of it. There's a bunch of like connector things just on different, like, yeah.
Marc:You know, they had to have some sort of inventory switch connected to a host machine. Yeah, you definitely know that. And you know what? It was probably stupid vampire taps and 10 base 2. Yep, yep, yep, yep. Yeah. Periodic polling over ARPANET. For the first time, you could query a physical object the same way you'd query a database.
Renee:Queryable matter sounds like the sequel to Jurassic Park.
Marc:Well, I mean, maybe it kind of is, right? Before that, the Internet was just about abstract data. Files, email, remote terminals. The Coke machine created a live digital representation of an actual physical thing. Like, even, okay, even the language of the, remember the old printer versus print device in Novell, right? There's this sort of abstract representation, right?
Renee:Yeah, it happened.
Marc:So you got this continuous state feed, you know, for this Coke machine here. Is it there? Is it cold? And it gets updated in real time.
Renee:So the hallway had a machine. The network had its ghost. It's the ghost of the machine, right?
Marc:A network visible ghost, though, right? And once you can query the ghost, you don't need to walk down the hallway. The network already knows.
Renee:So the Internet of Things began as the Internet of Avoiding Mild Sadness.
Marc:That's where it all started.
Renee:Mildly Irritating Things became the Internet of Things. That's where it all started, y'all. In 1982, the halls of Carnegie Mellon University with graduate students who were too lazy to walk down the hall.
Marc:Thanks, you guys. The Internet of Caffeine. The Internet of Cold Caffeine.
Renee:Let's fast forward to the early 1990s and Xerox PARC. Beige workstations, whiteboards full of diagrams that look like the future trying to sketch itself, right? Mark Weiser introduces the concept of ubiquitous computing. And what's important is what he didn't imagine. He didn't imagine phones glued to our palms, notification storms or glowing rectangles competing for our dopamine. The dream wasn't screens everywhere. It was invisible computing, right? Technology woven so seamlessly into daily life that you barely noticed it. Oh, like neural link, like it's in your brain, but you don't know it's there. Computation that recedes into the background like electricity or plumbing. Calm technology, a world where information surrounds you, but it doesn't demand anything of you. What a wonderful thought.
Marc:It's a wonderful thought. I read that paper, his first paper, and it comes out in like 1988. And it takes him a couple of years of this kind of idea, developing it. And, you know, it's just this, you know, by the 90s, the early 90s, 91, somewhere in there, they're like, it's a fully fledged idea, right? It's a philosophy. So, you know, what we've built instead, though, right, is sort of this anxiety machine. You know, we have ring-xiety, right?
Renee:Yes, I have ring-xiety for sure.
Marc:Yeah, phantom ring syndrome, right? But, you know, kind of you look at the architecture, Weiser was describing something very specific, an environment saturated with microprocessors. And, you know, if you think about it, there's microprocessors in everything today. Small network computational units embedded into objects and into spaces. Thousands and thousands of tiny computers inside of one big one. You stop going to the computer, the computer is already everywhere.
Renee:The center of gravity moved. Computing stopped being a destination and it became a condition. And while consumers didn't notice it immediately, the infrastructure was evolving quietly. So RFID tags became cheap enough to attach to pallets, cartons, even individual products, right? Like RFID is part of your passport now, right? Like there's tags in there now. um suddenly supply chains weren't just imagines they were they were sensed they were sensed like you knew where the pallet was because the rf rfid controller knew where they were you could watch stuff move around the warehouse right amazing stuff.
Marc:Well you know just thinking about rfid those are those are chips you know those are those are tiny microchips you know and they're not on all the time they require the field to energize them right for the for any compute to happen but like that's that's really chips everywhere right you know they're in everything and and rfid doesn't doesn't really get enough credit here each tag is the little little you know kind of controller if you look at i don't have one here and nobody would see it anyways But your smart card, right? That little chip on there, that's like some of the tags are different than that. But that's what you're looking at, right? And it's got an antenna and it gets wrapped around inside like your credit card and stuff. But, you know, it doesn't have a battery. It doesn't have any kind of plug or anything like that. You know, it doesn't have like line of sight, no manual barcode scan. The objects themselves become trackable as they move through space. And, you know, the energy of an RF field, you know, a radio field, energizes the chip. And then it sends its, you know, its information out. You don't count your inventory and reconcile. You detect it and then confirm it. The warehouse becomes a live sensor environment. I remember Walmart making this, like... Like, if you didn't have RFID technology, you didn't do business at Walmart. End of story. Oh, wow. Like, that was a huge change for them. And a lot of vendors were like, ooh, this is really hard. But if you wanted to do business with Walmart, you, you know, had an office in Bentonville and you had tags on everything you sold.
Renee:Wow. So the world starts labeling itself. Well, at least the vendors at Walmart did, right? Right. Boxes gain identities. Pallets gain histories. Products gain movement trails. The physical world begins accumulating tons and tons of metadata.
Marc:Oh, precious, lovely, juicy metadata. You know, when those physical objects have unique identifiers and pathways, they become addressable as well. You can query them. You can locate them. And Renee's favorite, you can audit them.
Renee:Audit them. It is my favorite.
Marc:Every tagged object is an entry in a distributed database somewhere that happens to have mass instead of just bits and bytes.
Renee:We didn't just connect computers. We started tagging reality. And that's the subtle turning point, right? Because once reality has tags, reality becomes searchable.
Marc:That's terrifying.
Renee:That's terrifying, right? Like, here's how I think about smart thermostats. Like, like that's a, that's a network, that's a network, right? And if you wanted to hack it, you could. And every node on that network can now do whatever that hacker wants. That's craziness. It's craziness. That's crazy.
Marc:Well, that, that thermostat examples is neat because, you know, there's temperature sensors and all sorts of things in your house. Right. And that's the whole concept of a smart home, right, is that these little sensors live in all sorts of different things. And these apps bring all of that, that metadata, all of that information into, you know, into one place. And, you know, you can adjust it.
Renee:Interesting you should say that because the Ring camera, like I think people think the Ring camera is just looking like, oh, I see who's there on my phone. I'll answer it or I won't. Actually, no. It's collecting so much environmental data that if you go to AccuWeather and it shows you hyper local weather for your street, where the hell do you think it's getting that from? Oh, wait, it's getting it from the Ring cameras on your street. Like, actually, Ring sells AccuWeather that data so they could have hyperlocal data in near real time. So, yeah, it's crazy. It's crazy what those proprietary networks are, what they collect, and who they sell that data to. It's pretty amazing.
Marc:Episode two, hyperlocal weather.
Renee:Hyperlocal weather.
Marc:So, look, once that data is searchable, right, all of that information, that metadata, you can also optimize it. And, you know, Weiser sensed this, but didn't quite get it there. He was thinking about this kind of more, I would say, more idealistic, you know, kind of community of compute. He talked about calm technology, computation fading into the background. The logical extension is computation embedded into the physical structure of the world itself.
Renee:Do you think he was disappointed? Yeah. Do you think he's like, oh, too bad. You guys misunderstood what I was saying. I think so. I was going to, like. Probably didn't care, actually. We didn't care what you had to say, but because I think that's the way it goes. Maybe. Maybe he's disappointed.
Marc:I didn't have time to research him on that one, but, yeah, it'd be interesting to check out, like, he's probably some grumpy old dude, you know, like, really angry. Right?
Renee:Oh, we should have him on the podcast being like, I never meant. For everything in the world to be talking to everything else, right? But I swear to God, if we didn't have sensors and pipelines, right? If we didn't have IoT, we would be doing a lot more work and we wouldn't be as fast at it. So I get the dream, but I think the reality of a noisy world is just noisy technology. So let's talk about the cloud. Early IoT experiments were constrained to physics and money, like most things, physics and money. Storage was expensive, Compute was expensive. Bandwidth was precious. Wait, before I go on any further, have you seen the movie BlackBerry on Netflix?
Marc:Yes. Yeah, I told you about that.
Renee:I watched it last night. Yeah, I watched it last night. And when they were sitting there going, we can't fit more than 5,000 devices onto the Verizon network. And they're like, we've got to figure this out. We need more bandwidth. And what are you going to do, build more towers? Like, we need more bandwidth. And so, yeah. So bandwidth was precious, especially back then. If you wanted to collect data from devices at scale, you needed racks of servers, cooling systems, capital expenditure approvals, and someone in a hoodie swapping out hard drives at 2 a.m. Oh, do you remember when that used to be us?
Marc:Yeah. We were that person. Yeah. We were that people. We resembled that remark.
Renee:Yeah, we did. The dream of connected objects existed. The infrastructure to sustain them actually didn't yet exist. Then cloud computing explodes. And just one more reminder that the cloud is just someone else's computer. Just want to put that out there. It's not real. It's just someone else's computer. And suddenly the whole ceiling lifted. Like what we could do in the cloud, we could never do with the constraints we had in our own world, for sure.
Marc:Yeah. You know, AWS launches in 2006, right? Elastic storage, elastic compute. Before the cloud, if you wanted to collect telemetry from 10,000 devices, you had to provision enough servers to handle peak load and hope you guessed correctly, right? After the cloud, you just don't guess, right? You build it into your application. You spin up resources dynamically. You store the data, you know, cheaply. You scale horizontally instead of vertically. And you can stream telemetry from thousands, then millions of devices into, and now billions. Billions. You know, with handsets and things. Millions of devices into centralized systems without building a data center in your garage.
Renee:So the friction disappeared, right? You don't have to justify collecting the data anymore. You just collect it. Like, by default, you can do it. So we do. So we do.
Marc:Yeah. Okay. Was it last week or maybe a week before we talked about Java and memory, like how Java turned like developers into, you know, sort of sloppy. It's like, oh, garbage collection, it'll collect it, you know. Right.
Renee:Yeah. We can do whatever we want. Who cares if it's bloated and lame?
Marc:Yeah, nobody cares.
Renee:And look how we used to build servers, right? Like they would come back to you after doing load testing and be like, okay, we can fit 30 people on this server. Yeah, but at peak, we need it for 100,000 people. so how many servers do i have to build like like it was crazy how that's how we thought about stuff.
Marc:Yeah not maybe maybe
Renee:That code should be better so we can fit 300 on a server no no no 30 right yeah craziness it was crazy peak.
Marc:Versus average but like i think about that and you know it's the same thing whether it's whether it's you know memory it's compute it's storage you know all of these sort of cheap storage, cheap compute changes how you think about the device data itself. You don't have to shrink the data down. You know, if you look at like ISO 8583, you know, that was a protocol written in literally 1983. And it's the whole TLV tag length value, right? It's like zero, zero, zero, one, 53, you know, and all, and it's all compressed into that, you know, into that format. But now it's like a tag, you know, slash, you know, I'm going to name some random tags, you know, close that my blob of data that might be, you know, several hundred K like, and you're doing that over across millions of devices, you know, just it becomes it's cheap because the storage is cheap and computer is cheap. But, man, I don't know. Yeah. Anyways, you stop treating it as sort of occasional reports. There's this continuous stream of really fat data.
Renee:Yeah, right. Yeah.
Marc:Temperature readings every minute, you know, or every second, if you want. Motion detectors every second. Engine diagnostics every mile. You know, oh my gosh. Have you ever seen the F1 guys after a race? How much data they've got?
Renee:Oh, yeah. There's so much telemetry in those cars. Like it's, yeah, F1 cars are unbelievable.
Marc:Petabytes of data just from telemetry. The data is always flowing.
Renee:Yeah. So thermostats aren't just thermostats anymore. And F1 cars aren't really F1 cars anymore, right? They are highly monitored, you know, with tons of telemetry and real strategy comes out of that data set.
Marc:A data center on wheels that goes fast.
Renee:Yeah.
Marc:Yeah. Like, you know, the temperature nodes, they're like, it's a global data set. One thermostat says it's, you know, it's 68 in freedom units. That's, you know, 17 in normal units in my living room. You gave me the look. 10 million thermostats
Renee:Oh so like now that you live in the UK like Fahrenheit's not good enough anymore no.
Marc:It was never good enough like come on But Fahrenheit is a weird, like, there's the meme that, you know, with Homer, right? And he says it makes sense because it's always, it goes up as it gets hotter. Like, yeah, yeah, okay. But it's not tied to anything. Like, you know.
Renee:I don't understand why when we're like 44 below, they're like, yeah, don't bother. It's all the same. Like, well, why? Why is it all the same at 40 below? Like, I don't understand that either. Does some magic physics happen? Anyway, go ahead.
Marc:Well, yeah, your brain can't tell the difference anymore. It's just, oh, that's cold, man.
Renee:Right.
Marc:Yeah. Anyways, so a bunch of thermostats, they all work in conjunction. Here's how an entire region heats its homes in winter. Oh, you know what? I've got the Nest, right? And I don't have air conditioning because I live in the UK, right? But I do have a heater, a boiler. And it'll say, like, you were in the top 25% of users for the month of whatever. And I'm like, oh, did I consent to that?
Renee:Mind your own business. That's what I would say. The hell are you watching me for? You want everybody to freeze? What the hell?
Marc:Yeah, but, you know, what's interesting is that you collect all that information. You get things like occupancy behavior, right? Nest is a great example of an IoT device. You know, home, away, you know, vacation has these sort of occupancy behaviors. You know, where is your peak demand? And, you know, what kind of insulation efficiency or inefficiency you might need to look at. And, you know, if you look at it at scale, the individual reading is, it doesn't matter, right? So what, my house is, you know, at 18 degrees or whatever. But the aggregate, when you pull a lot of that stuff together, it's a real behavioral map.
Renee:I hate that idea. I just, I hate, so I hate smart meters. Don't even get me started on smart meter.
Marc:I know, I know.
Renee:So I had a dumb meter. And then so PG&E said, you can keep your stupid dumb meter, but you're going to pay us $75 a year for the, I have to give up personally to know that. I find it really awful. Awful. Anyway, we go from personal convenience to population level modeling. And if that's what we used it for, sure. I don't think that's all we use it for, though. And you can call me paranoid, but I think that we don't use it just for that.
Marc:Well, I think when did we talk? Oh, I think it was in the hyperlocal sensor one, right? We talked about millions of sensors being installed, right? What happens when you install tons and tons and tons of sensors across multiple sort of measurement domains is you get very, very accurate cross sections of information, right? Location, temperature, humidity, electricity, water usage, all of that stuff. When you start putting it together, you could absolutely create a profile of activity in a very, you know, focused area.
Renee:Yep.
Marc:Yeah, that aggregate telemetry is what makes, you know, kind of the machine learning practical. You can train models on them. You can predict energy usage. You can optimize load balancing. You can anticipate failures, cluster behavior patterns. You know, all of that is true. It's, yes, all of it's true, but it's like.
Renee:That's the benign way of looking at it.
Marc:I know. That's the benign way of looking at it. You know, but none of that works with the small data sets from individual devices. But as you scale that information up, it requires all of that, you know, cloud scale aggregation. Once you have that, the AI layers is sort of almost a given, right? And you can derive so much intelligence out of that information. And boy, it's a lot. And it's not anonymous. Don't even buy into the, no, it's not anonymous.
Renee:No, it's not. Nothing ever is.
Marc:Well, what is it you can figure out an individual from a zip code, like four digits of their social, and that's it, I think?
Renee:That's it. Yeah, that's all it takes. Yeah. Well, AI can infer all kinds of things about you. It doesn't really need to know anything. So that's what it is, right? So we go from, you know, objects become sensors, the cloud becomes the memory, and AI becomes the interpretation layer. Like, what's it all mean? AI is gonna tell us what it all means, right?
Marc:The value sort of migrates over time, though, right? So, you know, people get fascinated by the device, right? Everybody starts talking about the device. The smart thermostat, like the Nest. We just talked about my Nest, right? Connected watch. Oh, my gosh. Do you know how many, like, medical sensor testing kinds of, you know, things that your Apple Watch can do? It's insane. It can do hundreds of different types of medical diagnostic tests from your Apple Watch. It's nuts. You know, the different sensors, you know, but as that data starts to accumulate, the hardware becomes the collection mechanism. It's less sort of novel. And the data set is what the business is built on, right?
Renee:The device is just an excuse. Like, oh, I gave you this and now I know everything about you.
Marc:Yeah.
Renee:That's the point. Like, I think in that scenario, you should give me that watch. I shouldn't pay for that. Like, if you're making money off the back end of all that data, you should just give it to me. Am I accepting it means I'm okay with you taking all my data. But I guess my problem is I paid for it, and I wish you didn't have it because you charged me $700 for it, right? Like, I think at some point I shouldn't have to do that. We're just going to make me mad again, but okay, let's just keep going. I'm going to need a drink after this, too.
Marc:Yeah, yeah. So the heating behavior model, right, the usage curve, aggregated demand signal, that's what gets monetized. The device collects it, right, and the business gets built off the back of the data.
Renee:So we didn't just give objects connectivity. We gave them memory. And once the world has memory, it can be analyzed.
Marc:Yep. And once you analyze it, it can be optimized. At some point, I know, I know, optimization starts running in both directions, though.
Renee:Oh, okay. So let's just fast forward to 2010. And the infrastructure's ready. Cloud's cheap. Wi-Fi's everywhere. Smartphones have trained us to live inside apps. And that's when IoT crosses, the Internet of Things, crosses a psychological threshold. It moves from factories and supply chains to the living room. Like we've been saying, smart thermostats, smart speakers, smart locks, smart doorbells, smart light bulbs. Yeah, that need firmware updates. It's not just good enough to have a light bulb. I got to run a ROM update on that bad boy so that it'll keep working. We didn't just connect industry. We connected the kitchen.
Marc:The kitchen why i i was trying to find out i was looking to see like if i could find some stats on how much like how many chips intel and amd and arm and all these guys produced just for coffee machines and refrigerators and i couldn't get any concrete numbers but it like your car right like Like there's so many sensors and chips inside of every device that you see. It dwarfs the number of CPUs in a data center. Yeah, it's nuts.
Renee:And that's why whenever we talk about like Taiwan has a drought and we can't make any chips. So it happened, right? So Toyota is the only car company that kind of saw it coming because, you know, they're in that part of the world too. And so they knew they had droughts. So it was clear Taiwan had them. And Taiwan, like, actually... Plans on hurricanes to drop water right that's they plan on it and when they don't get them for a couple of years it's really detrimental and so like where toyota could get that stuff, ford couldn't so ford decides all right man we don't know what to do we can't you can't do a heads-up display in a lincoln that without it right like you just there's a lot you can't do in a car if you don't have the right chips as they decided we're going analog and people thought it It was like, oh, look, look, the little needles are back. And it's just like it used to be. Like, yay. And it's like, no, dude, they just couldn't get the chips. Like, they knew they couldn't get the chips. And so we were all like, oh, it's retro. Oh, look, my speedometer has a needle now. That's retro. No, they would have rather have not done that, except that they couldn't get any chips to do it any other way. So it has real impact, right? It has real impact.
Marc:Speaking of cars, have you seen this? I can't remember. I think it was lexus has started putting camera sensors inside the dash and like you know you're on your steering wheel there's like if you look at the you know your speedo and you know your gauges and stuff it's this little tiny thing and it's like the top of your iphone you know it's got a little pinhole or something and there's literally like a pinhole camera to see if you're no i know i know and and yeah they're starting to hide them in you know in the mirror and if you knew how
Renee:Much i sang in the car like i find that like i sing so much in the car it's the one place where i can sing really loud and no one yells at me like i it's terrible.
Marc:But it's terrible it is terrible but like the 2010s though like to me you know cheap semiconductors cheap storage cheap compute the other thing cheap wireless right cheap wi-fi but then the advent of 4g and 5g like like nothing like supercharges iot but you know except 5g it's this you know huge like enabler for everything and yeah yeah
Renee:Remember the promise of 5g was that But a surgeon could remotely do.
Marc:Surgery with a robot.
Renee:Yeah, and there would be no latency. And it was just all wireless, right?
Marc:Yeah. No, but all your little devices, they're phoning home and talking. Now I'm starting to sound like a conspiracy theorist.
Renee:No, you know, Dick Cheney, he had something in his heart where every night he would lay down. Well, it was clearly a pacemaker or something. No, no, it was a pump. So, you know, he had, okay, so he had heart failure, like full-on heart failure. His heart did not pump. So he had a pump in his body that would just like circulate his blood for him. He did not have a beating heart. And every night when he laid down to go to bed, it would take all the telemetry data, push it over to this device here, and then that data, because it's 5G, that data would go all to his doctors, cardiologists, and then it would run it through AI. And if it popped up anything, Dick, you got to come in. Or if it didn't, Dick, you don't have to come in. Like, that's how it worked. And so he spent years having no heart.
Marc:That's
Renee:All i'm gonna say.
Marc:Years all right let's let's bring it back to like homes and houses and sensors and stuff right homes become local sensor networks right we're talking about the 2010s here each device acts as an edge node like you you have the home app on your on your iphone or whatever it doesn't yeah i do yeah you got the alexa right you talked to alexa no i unplugged Oh, you unplugged her? Well, that's probably a good point.
Renee:So I did an experiment where I said Scout. Instead of saying Scout, because Scout was the dog, I would say Eggplant. And then it figured, like, and then it's Christmas time and I open Amazon and it's like, Eggplant needs a sweater. I'm like, that's it. That's it. I unplugged all of them. When it inferred that Eggplant met my dog like that and it was trying to sell me stuff, I unplugged all of them. They're all sitting around. They're just not plugged in anymore. I am paranoid.
Marc:But a smart home, you've got all this stuff running around. It captures all of these different signals locally, audio, motion, temperature, occupancy, does some lightweight preprocessing. And then it ships that structured telemetry to cloud-based inference engines. Your smart speaker doesn't know anything in isolation, right? Alexa doesn't know anything about eggplant. It infers it later on down the road, right, the data when it becomes inferred. It's basically just a microphone or Wi-Fi and a subscription to hyperscale AI.
Renee:So now my living room is a peripheral device.
Marc:That's so depressing. Your house is a distributed input system for remote models, Renee. The architecture is simple. Low power hardware at the edge. High power compute in the cloud. Constant connectivity between them.
Renee:And we used to talk about home as a sanctuary. Now it's just a mesh network with throw pillows. It's depressing. It's depressing.
Marc:We sound old. We sound old like the... I guess... I guess for you and I, it's not so much that we're sound old. It's that, like, we know better about privacy, you know, than some people.
Renee:And we didn't stop it. I think it continually, like, depresses me that we were part of all of this, right? Like, we started it. Like, we started it. And we didn't do anything as it was coming up. Because every time something came up, we thought technology was benign. It's benign. It's benign. Look what this is going to do for us. Look how happy we're going to be. Oh, look how great this is. And then every single time it's monetized on the back end and it's eventually weaponized against us. And so, like, I don't know why we keep falling for it. Like, AI is benign.
Marc:No.
Renee:This time I'm not falling for it. It is not benign.
Marc:It's not. Oh, geez. Yeah.
Renee:All right. Go ahead.
Marc:So before IoT, the home was reactive to you physically, right? You flip a switch. The light turns on. You know, now you've got apps and controls and all that stuff. If your home can anticipate what you're doing, this should be good, right? Your home anticipates your needs, Renee. Motion sensors detect your presence. Thermostats learn arrival patterns. They optimize energy usage, right? Doorbells trigger phone alerts. Speakers respond to your voice commands. Feedback loops everywhere. It's Wiser's vision, right? That, you know, it's all working for you.
Renee:We gave the walls a nervous system. I'm not buying this. I'm not.
Marc:It's not even metaphorical, though, right? You've got motion detection, audio input, camera feeds, environmental sensing, all feeding into some, you know, decision logic. The control systems theory, it's a closed loop. Input, processing, actuation. Your house is now a control system.
Renee:Which I get, okay, if I'm being fair, it's sort of fascinating, but it's really unsettling because we didn't frame it that way. We framed it as convenience. Never touch a thermostat again. Unlock your door from the grocery store. Like, why would you even do that? Why is that a good thing? Like, you know, ah, ah, ah. Ask the weather, ask for the weather out loud. But structurally, what we did was embed continuous monitoring into our domestic life. I can't even stand continuous monitoring in my professional life. And now it's in my, it's so bad. Here's what it is. It's so bad that my watch, if I'm a little unsteady on my feet, because I'm hard of hearing, it causes me, sometimes it causes me balance problems. And so every once in a while, my watch will be like, Renee, you're going to fall in the next two weeks.
Marc:What?
Renee:What the hell? Number one, why are you bringing it up? You're just going to make it happen. You're putting it out into the universe, so it's definitely going to happen, right? And now you got me freaked out about it. Why is this a thing, right? And so, yeah, I don't want to be continually monitored in my own home. Like, that should be the one place where I'm not subject to an algorithm. Like, do I get no peace? No.
Marc:No. No.
Renee:Yeah, no.
Marc:I had to get rid of the iPhone, the iWatch, you know, the Apple Watch, man. I just, I couldn't. It tells me all the time,
Renee:Renee, you're going to fall. Like, why are you bringing it up? Stop it. Stop it. I mean, maybe it would be helpful if you said you're going to have a stroke, but no, no, I'm just going to trip over something.
Marc:I couldn't deal with the constant buzzing on my wrist. I don't know. It was very difficult to tune all of the stuff, all the alerts and everything. If you've got like 100 apps on your phone and they all want to buzz you and stuff, it's like, go away.
Renee:Yeah, yeah, I hear you. I still wear mine, but I wear it for the exercise crap and the heartbeat crap.
Marc:Yeah, yeah, that's cool. Yeah, yeah. But, you know, the smart home... Okay, so we talked about 5G, talked about Wi-Fi, talked about compute, talked about all these things. Okay, APIs, massive explosion, right? Open developer platforms allow third-party services to integrate to your smart home, to your IoT. Lights can talk to speakers. Locks talk to cameras. Thermostats talk to utility companies. Ah, no. Interoperability turns scaled devices into ecosystems.
Renee:So now my light bulb is socially connected. It has its own stupid Facebook page. I'm not happy with this. I'm not. Go ahead.
Marc:Yes, your light bulb has friends, Renee. Oh, my God. And once these devices interoperate, the system gains the complexity. You start orchestrating behavior across them. If motion is detected after 10 p.m., turn the hallway light on at 30% brightness. You know, it just seems like really complicated. Just turn the freaking light on. If your thermostat detects absence for 48 hours, shift to energy saving mode. If the doorbell detects motion and face recognition matches a known visitor, send a push notification with contextual metadata. Like, you could orchestrate all of it.
Renee:Just open the door for them. Yeah, just open the door for them. What do you care? This is crazy. Are we that lazy we can't flip a light switch? You know what? Yeah, it's the logical extension of I'm too lazy to see if the Coke machine is empty. I mean, this is where we end up, right? Yes.
Marc:Yes, yes.
Renee:We built a house to narrate itself. It's just, it's ridiculous.
Marc:Why is there one in calm computing, right? What we actually built was notification computing. All of these devices are constantly servicing themselves. The firmware updates required, the camera detects motion, subscription expired. Oh, okay. I bought the damn, you know, camera. I bought the damn, you know, thermostat. and I've got to pay to use it. Every device wants your attention.
Renee:Yeah. The refrigerator doesn't just chill milk. It emails you. I don't need more email. And I sure as hell don't need it from my refrigerator. I can't even think of a moment where that's a good idea. Like, hey, Renee, you're out of milk. I know. I know. Are you going to order it for me? You're going to go to Amazon, order it, have it delivered, and then have it just show up? Because maybe that'd be okay. But just telling me I don't have milk, you're not helping. That doesn't help.
Marc:I think I can build you a system that does that. Can you go to Claude? No, no. I already have the APIs and everything. That's not a problem if it wants to order your, you know, that's agentic commerce. That's a perfect use case for agentic commerce.
Renee:Yeah, my smart refrigerator knows what it's out of. It adds it to the cart and orders it.
Marc:And that's a use case that's been around for a while, right? So it's not like it's, you know, it's a brand new one. But, you know, it's always been a little bit clunky. And I think that some of the orchestration, the metadata, you know, all of that stuff, I think it's getting better. But, you know, your refrigerator. But you know what? I don't want to pay. I don't want to pay a subscription to use the thing that I bought. I hate that.
Renee:I agree. Yeah. Again. Have you seen? We monetize everything.
Marc:I know. Have you seen the car companies? This. Okay. Spoiler. We'll save this for the car episode because it does. I. Yeah. The car episode. We'll talk about this. Paying for your, you know, car subscription stuff. Just. Yeah. Anyways, I'll get off of that. Anyways. So the refrigerator is collecting more data than you'd think, right? Usage patterns, door open frequency, internal temperature variance, compressor cycles. Compressor cycles might be a good one. Aggregate that across millions of units and you have behavioral analytics.
Renee:So the milk shelf is part of a data economy.
Marc:Yeah. Yeah, right. The device is the visible object. Behind it sits longitudinal behavioral data. Energy consumption patterns, occupancy signals, voice recordings.
Renee:So we invited sensors into the most intimate space in our lives, and we called it minimalism. This is why I won't get a smart toilet. I don't want it to know anything. I don't want it to be like, Renee, you might have colon cancer. No, it's none of your business. It's none of my business. I don't want to know anything. Don't tell me I don't care. Because I don't think for a second that's not going to end up in the hands of Mark Zuckerberg somehow.
Marc:I see what you're saying. I get it. But I kind of would be okay with my toilet telling me that.
Renee:Really? You'd be okay with that?
Marc:But only because, only if the health data was not, you know, aggregated and shared. Like, I know that's impossible, right? Like, that's the model. That's what will happen. But I'd be okay with, you know.
Renee:See, I'm okay with the water treatment facility running tests and being like, uh-oh, the zip code has COVID. Like, I'm okay with that. I don't know if it's okay with, hey, Renee, you know, you're eating too much sugar. Like, mind your own business.
Marc:Yeah, yeah, I get it. I get it. But heck, man, New York, they, you know, they banned the Big Gulp, you know, so. I mean, maybe. It's living well with me. There you go. So maybe, you know, Mom Donnie says smart toilets for everyone in New York, and that's how they bring down health costs. I don't know. But, you know, that exchange is incremental. It's like one device at a time. It didn't feel structural, right? You know, when you replace your refrigerator, that's one thing, right? You replace the microwave. You buy a car, right? Incremental change, one device at a time. It felt like you're buying gadgets. It didn't feel like a structural change in your life.
Renee:No, we didn't realize we were building a domestic operating system, right? And then selling it off to third parties. It definitely wasn't my intent, for sure.
Marc:Yeah, yeah. The home is software-enabled. It's surveillance, you know, and it runs on code.
Renee:I was going to talk about smart meters, but I already said I chained myself to it. I was really mad about it. So we just keep going. But again, if you have a dumb meter, you should pay whatever amount of money they want to keep it. That's my advice to you.
Marc:You know, the guys come, like, every six weeks or so because they can't find the water meter at our house. Oh. And, like, the water meter is somewhere on the property. Like, even the previous owners, they're like, we don't know where it is. So the guys show up. They're like, we're from the water company. We need to take a meter reading. We're like, good luck.
Renee:Good luck. I don't know where it is.
Marc:I don't know where it is.
Renee:Did you bring a metal detector? Head out to the yard.
Marc:Yeah. Well, they go out, they check all the places that I think that there's something, and they're like, yeah, but can't find it either. I'm like, well.
Renee:So was your water free, or did they just charge you, like?
Marc:No, they charge us a base kind of unit, sort of, which is fine because there's six of us, and the pool, and the dog, and, you know, so,
Renee:Yeah. Yeah, okay, that's fair. So, okay, so consumer IoT is adorable. Your fridge sends you a notification. Your doorbell records a raccoon. or the cops. I don't know. Your light bulb has opinions. Industrial IoT does not send you push alerts. Industrial IoT keeps the lights on. It doesn't make headlines. It rewires infrastructures. Factories embed vibration sensors deep inside turbines spinning at thousands of RPMs. Oil rigs monitor pressure fluctuations miles below the ocean surface. Wind farms stream rotational telemetry from blades the size of skyscrapers. Agriculture deploys soil moisture sensors across thousands of acres. Thousands of acres. And they monitor it from Germany sometimes. It's crazy. To optimize irrigation down to the inch. This is not convenience. This is civilization quietly doing. It's instrumentating itself, right? It's this idea that with more data, I can better manage industrial stuff. And with this one, I agree, right? Get weird telemetry from a windmill and you have an opportunity to shut that thing down, fix it before it blows apart and breaks other windmills, right? Or kills people. That's actually really important stuff. So I don't mind industrial IoT. I actually think that's a really good thing.
Marc:Yeah, I'm a big fan of the industrial IoT. You know, predictive maintenance changed a lot of the economics around big industrial installations. For most of industrial history, systems are reactive, right? A bearing fails, the turbine seizes, a pump overheats, then you fix it. But once you embed sensors, you can monitor micro anomalies, you know, one and a half percent deviation and vibration frequency, right? A subtle shift in harmonic resonance, two degree variance and thermal profile. Those are signals. Remember the dot com? I literally could tell you how many revolutions of the fan was making on the CPUs on every single server. Yeah. Yeah. It's collecting all of that information continuously and you can train models to predict failure before it happens.
Renee:Which, if you want to go back to the data center, like, imagine being one of those hyperscale data centers. Like, it will tell you, hey, this fan's going to fail in about, what, three weeks? You might want to schedule a replacement for it now. Or this CPU is coming to its end of life, you know, you might want to replace it now. I mean, all of that stuff keeps you, on the positive side of that equation, not running to fix things after they break. And yeah, so instead of machines breaking, machines whisper. They emit faint signals. I know, right? Like, help me. They emit faint signals of degradation long before the visible collapse. And when you aggregate that data across fleets, hundreds of turbines, thousands of pumps, you build probabilistic failure curves. And that becomes really important to maintenance engineers who have to keep all that stuff running.
Marc:Yeah, you schedule the maintenance by whatever condition, you know, instead of by the calendar. The machine tells you when it needs the attention.
Renee:So machines gain, I hate to say this word, conditional awareness, but they kind of do. Like they're not aware, but we're aware of what their patterns are so that we can put that in the right context and do something with it.
Marc:Yeah, sure. It's state awareness, right? You can report, you know, they can report on their own condition. And, you know, if you have that continuous telemetry, you can build your digital twin. A digital twin, right? Yeah, it's cool. It's a virtual replica of a physical system. You continuously update with real-time data. You run a simulation of that turbine in parallel with a real one.
Renee:So the factory has a shadow factory. I love digital twin. I think it's a really cool thing.
Marc:It's a cool thing. Yeah, it's a cool system, right? Physical system operates in the real world. The digital twin models stress loads. You can change temperatures, temperature gradients, material fatigue, performance curves. You can do all of that in real time because you have all that telemetry data. The shadow can run scenarios faster than the physical system actually experiences them.
Renee:It's like the building has a parallel life. And if you pair that data with, like, I don't know, sustainability data, climate risk data, and say, yeah, at this time, you know, 10 years from now, there's a high likelihood that we're going to see rainfall that's going to accumulate and it's going to be, like, more than 10 inches. Like, let's put that at this building. Let's throw 10 inches of water at this building and see what happens to it. And then you start to realize, uh-oh, we need to get, you know, stuff off the floors. Maybe the data center should be on the fifth floor instead of the basement. Like, you start really thinking about that stuff because two things are true. Climate change is real and that stuff breaks. So, yeah, I think it's all super important. I love that stuff.
Marc:Who puts a data center in a basement these days? I don't know. Sometimes the shadow, you know, the shadow system will know bad things happen first because it can see the trends that humans can't.
Renee:So industrial IoT isn't about, you know, smart factories. It's about anticipatory factories. And this extends beyond manufacturing. Smart grids balance energy demand dynamically. Power plants modulate output based on real-time consumption models.
Marc:Yeah, but you can, you know, rail systems monitor axle stress and track integrity continuously. Water systems detect micro leaks before they become bursts.
Renee:Oh, you know, Hollywood could use that. Yeah. All the time, Hollywood will have a water main break. It happens constantly over there.
Marc:Okay. This is a bad stat, but I think for the UK the last year, over a trillion liters of water leaked. What? Yeah. Yeah.
Renee:That's really old infrastructure. I mean, you guys have infrastructure that's like hundreds of years old.
Marc:Yeah, I know. But that doesn't excuse it, right? Right, no, it does not. It doesn't excuse it. Just, ugh. The amount, like, okay, they paid themselves, the UK water industry, because it went private, they paid themselves $85 billion in, you know, rebates and, you know, buybacks and bonuses and all that, you know, kind of stuff. And guess how much they're in debt? About $85 billion in debt. You know, it's like... Excuse me well
Renee:You're a citizen go complain to the crowd.
Marc:I know now now i now i will see now i really sound like you know get off my lawn yeah
Renee:It's okay we've earned it we've earned get off my lawn yeah i'm afraid yeah yeah.
Marc:You know the power grids monitor load at the transformer level which is that's a great yeah that's a great benefit yeah
Renee:It's like infrastructure grew nerve endings. It's really cool.
Marc:Yeah. And once your infrastructure has that sensing layer, it can adapt. Steel and concrete with embedded feedback, bridge reports, its own stress load. Yeah, definitely need that.
Renee:Yeah. So while we were arguing about smart speakers mishearing us, which I argue about all the time, civilization was quietly installing a nervous system under the hood.
Marc:Yeah. I mean, other than nerdy people like us, nobody talks about industrial IoT the way They talk about smart speakers, but the economic impact is orders of magnitude larger. The downtime decreases, waste decreases, efficiency increases, compounds, you know, significantly.
Renee:If civilization can now sense stress in our own machinery before failure, what happens when we start sensing stress in ourselves the same way? What do you think is going to happen there?
Marc:I don't know. Like, we just talked about the health stuff. I mean, do you
Renee:Think that— Could you make a digital twin of me? Like, I wonder when we're going to get to the point where medical trials aren't really medical trials anymore. You create digital twins of, like, different kinds of people. Men in their 50s with heart conditions, women in their 30s, you know, with Lyme disease. And then you just throw conditions at that. And then you say, all right, well, we have a car ball of just stuff. And this is what the med looks like. Let's throw that at it too and see if it cures any of you, right? Like, will we ever get to a point where we'll have enough sensor data of real people that we can anonymize and then use that for virtual medical trials. Like, I have to think that, why not? Like, why not? Like, why not? Like, I just feel like there's huge potential in all of that. And maybe you get drugs to market faster. Maybe you are able to develop drugs for, like, small numbers of people who have really rare diseases. Like maybe we can cure a really rare gallbladder cancer that can have someone live longer. Like I want that. And strangely enough... Oracle stock went up by like some crazy amount, like I want to say 30% when Ellison just said that out loud. He said out loud, he's like, yeah, why not? Why can't we use AI to create individual cancer treatments for everybody on Earth? And with that, his stock went up and made him like almost a trillionaire. And then everybody realized you can't do that with a Gen AI model and it crashed again. So, you know what I mean? So, like, I think there's a huge amount of potential there. I just hope we don't squander it.
Marc:But there's, you know, that's a, I mean, that's why Jobs, Steve Jobs, you know, kind of first bout with cancer, he was able to survive. Because he spent the money to basically sequence himself. And then they used that information to create treatments that would work for him. And, you know, he had the money to do it. So, you know, okay, which is, I'm not sure where the ethical boundary is on that, but I mean, the science, the science needs to be done. He was paying for it. Okay, great. He benefited, you know, it didn't last obviously, but I just think that's, you know, that's definitely a potential there. And I think, you know, as compute has gone up, you know, sequencing and, you know, individual, the capability to individually sequence people. Yeah, it'll come. It'll definitely come.
Renee:Yeah. Okay. So then came the reckoning. The botnets, webcams, routers, DVRs, hijacked because they shipped with default passwords. Oh, the dreaded default password. Like when the administrator password is administrator, you're lame. Like the first thing you should do is change that thing. There's a reason for that.
Marc:Yeah, IoT massively expands the attack surface. Every device is a tiny computer, and every tiny computer is potential vulnerability. And because they're so tiny and the surface area is just sort of small and squishy, and you're going to deploy it to your kitchen countertop instead of into the data center, right there, they're soft targets. We optimize for all the speed, convenience, but not security architecture.
Renee:And we connected everything before we secured it. Any of it. And if I always thought like, if I were a nation state and I, and I didn't want to blow up the aquifer, right. I could just get your nests, you know, smart thermostat, turn it up to 150 degrees and let you suffer. Like you'd have to go sit outside. There'd be no way for you to turn it down. Like, and, or have your refrigerator just email you constantly, just constantly turn your, turn your washer on so often that you can't actually wash anything because it's always locked and it's always running. Like, as a hacker, I would just want to irritate the crap out of people. I wouldn't want to hurt anyone. I would just want to irritate the crap out of you. That would be my, like, why is she doing this? She just wants to irritate people.
Marc:Yeah. Well, I mean, it's sort of a classic tech move, right? Just, you know, just connect everything. But, you know, the IoT attack was, you know, what was it? 2008? One of the times that one of the Iranian nuclear enrichment centers went dark was because basically it was an IOT attack against their electrical infrastructure.
Renee:Oh, what was that called? Yeah. Oh, I know what you're talking about. Yeah, it made the centrifuges just spin until they exploded.
Marc:Yeah.
Renee:So they couldn't enrich anything. Yeah. What was that called?
Marc:That's an IOT hack, right? Yeah. That's an IOT hack.
Renee:But then once you do it, like now they know sources and methods, right? So you can't do it again. Yeah, that was crazy. Oh, and I think every time you see a rocket blow up in North Korea, like that's us. We definitely did that. We definitely had some jacked software in there that just, yeah, for sure. All right, so listen, now we're in phase two. Phase one was connection. Phase two is cognition. IoT without artificial intelligence.
Marc:Pseudo-cognition.
Renee:Right, pseudo-cognition. That's what else. It's fair enough. It's just telemetry. It tells you what happened. Temperature readings, motion events, pressure levels, door open, door closed. It's reporting. But IoT with AI, that's autonomy. That interpretation layered into signal. So the moment the system stops asking for instructions and starts generating them, a weird wobble on a turbine, just shut it off. Shut it off and tell somebody you did it. Like, that's what we're talking about when you layer AI onto the IoT.
Marc:Yeah. Sensors can feed models. Models can produce probabilities. Probabilities trigger actuations, right? Actuators modify environment. It's all closed loop. The human used to sit between the sensor reading and the response, but now you can build a model to do that for you.
Renee:So instead of waiting for me to adjust the thermostat, the system anticipates it and adjusts it. Oh, I wonder if it has a menopause mode, because I could have used that. I was having so many hot flashes. If I could have hooked my watch up to the thermostat so it knew I was having a hot flash and it would just turn the air on, That would have been helpful, actually. I might have liked that, right? Instead of waiting for a traffic engineer to reprogram signals, the grid recalculates it in real time. Instead of waiting for a building manager to respond to occupancy changes, HVAC systems reconfigure themselves dynamically.
Marc:Yeah, traffic signals optimize flow based on live congestion data and reinforcement of learning models. Smart grids balance energy load real-time, shifting supply based on consumption forecasts and renewable variability. Buildings can use predictive occupancy modeling to adjust lighting, air flow, and temperature before people even enter the space.
Renee:Like you can hook, like, I guess you could take like door card keys and at a certain point you're going to realize there's so many people in the place that you should actually adjust the temperature, right? Yeah, I mean, yeah, somebody doesn't have to go do that. In control theory, you move from open loop to closed loop adaptive system. Open loop, sense report, human decides. Closed loop, sense model, adjust, repeat. The model learns from each cycle. each intervention becomes training data for the next one. So maybe, you know, like it paid attention to the car door. There were 15 of us in there and it made it really cold in there. We're like, yeah, no, no, no, no, I'm going to change it. And so it's like, oh, you changed it. So the next time there's 15 people in here, I'm going to make it this warm, I guess. Yeah, that's smart.
Marc:Yeah. But, you know, that sort of optimization requires some sort of objective function, right? What do you want to do? What's the outcome that you want? Reduce congestion, minimize energy waste, maximize throughput. You can increase efficiency, right? Every system acts toward a defined goal. Someone has to choose what the goal actually is.
Renee:And whoever defines the goal defines the system's behavior. So if you leave temperature up to a menopausal woman, you're all in trouble. Bring a hat.
Marc:Yeah, the architecture encodes decision-making, right? And decisions can imply agency. even if that agency is bounded by the code itself.
Renee:So the Coke machine just... Important if the soda was cold. Now the system decides how traffic flows, how energy is distributed, how buildings breathe. We started by asking objects questions. Now objects answer on their own.
Marc:Scale change is what this means right millions of micro decisions happen every second across the infrastructure traffic lights grid switches climate systems the aggregate reshapes behavior you leave five minutes earlier because traffic adjusts you use less oh wouldn't that be nice like you know i'd be like oh look everybody wants to like the traffic engineers want you know a continuous flow rather than stop and go. So they alert, you know, people based on random assignment. Leave work a little earlier.
Renee:Oh, leave at 2.30.
Marc:Yeah.
Renee:It's like, oh, thank you. Thank you. I think I will.
Marc:Yeah. Oh, I'm giving myself permission to go. You know, use less power because your home throttles consumption. No, I'm not okay with that.
Renee:No, right?
Marc:Crank it up.
Renee:Yeah.
Marc:You walk through a city whose lighting and temperature subtly respond to occupancy patterns. I think that's cool. That's good. You know? Yeah. You move inside a responsive system, and most of the time you don't notice it, hopefully.
Renee:So if every object in your life is sensing, learning, and adjusting, are you living in a home or inside a constantly updating behavioral model of yourself? Again, how do I escape the algorithm? Like, I cannot escape the algorithm ever. Like, in a smart home, I'm just telemetry data. Again.
Marc:Yeah. Right? So I don't have the luxury of experiencing this because I live in an old house. So, like, the house we lived in when we moved to the UK, it was older than America, right? I mean, so it literally was built around the time, you know, that the Spanish Armada, you know, was, you know, pulling into ports in England. You know, it was literally like by the beacon fires, you know, that were lit. So it was not a smart home in any way, shape or form. And so now, you know, I have a Nest that's cool, but it only does, you know, the boiler. Everything else is, you know, kind of old systems and stuff like that. So, like, I don't have the luxury of knowing, so I'm not sure I miss it, you know?
Renee:Here's what's crazy, though. So there are hotels all over the world. Some of them are very old. But in this idea that I have to report on sustainability requirements, there's this huge push to retrofit that stuff into smart stuff. So like low water usage and reporting it, like and reporting it, right? So and then low electricity usage and reporting it. Like, so they found themselves having to take that same situation you have and figuring out a way to put enough sensors in an old building to be able to say, here's my scope one, my scope two. And so, and that's currently underway all over Europe. Eventually it'll come back to the United States. But yeah, I don't, I don't know. So I do feel like I'm just adding to my problem that I'm at the mercy of the algorithm. And the more you know. At least, you know what, at least when I was on the internet and you knew where I went and what stuff I looked at, what I was buying. But now you know what room I'm sitting in. You know, you know how long I'm there. You can see me walk around the house. You know when I take a bath every night. Like, at some point, I think you know too much. Like, my mother doesn't know that much about me. Like, you don't need to know that much. Like, I get to be anonymous. And if I'm wearing those stupid Ray-Ban glasses that stupid Meta makes now, like now all of that is streaming to someplace I don't know. And people are watching me take a bath. I don't know. Like it's borderline crazy to me. Like I don't think you need to know this much.
Marc:There you go. The glasses one is interesting. So here's a little smart tip for you and the listeners. There's an app. If I remember it, I'll put it in the show notes. There's an app that you can get now that detects when you're in the vicinity of somebody with the glass hole, you know, glasses.
Renee:Can I smack them? Or do I have to put a bag over my head? Like, what do I get? Is that standing around? It just warns you.
Marc:Yeah, no, it just warns you that somebody's around. And it's kind of clever because what it does is it reads the Bluetooth, you know, IMEI. That's not an IMEI. It's a UUID or something like that.
Renee:Yeah, you know what? I'm going to look it up and install it just to do it.
Marc:Just to be a joke. No, it's, it's, you know, it's a smart idea because, you know, the, the lack of consent in these models is, yeah, it's, it's atrocious. Right. And I think that's my issue with the smart home information is that like, it's not anonymous. Even if they say it's anonymous, like, a good data scientist could un-anonymize it, you know? Right. So, either we have to get better at actual anonymization, right, to decouple.
Renee:Don't collect it to begin with. Just don't collect it to begin with.
Marc:You know, decouple location from, you know, that usefulness. If you decouple that, then the anonymization problem is a lot easier to solve. But you know like i would pay more if i knew that my data was my own only you know and the model was trained only on my information like i would pay more because you know then i then i know i
Renee:Just it was my whole thing with my dumb meter i'd pay more for it just so you know it protected my data right like you don't know anything about me now you just know that.
Marc:Every couple months you know like there's way more there's way more money for google you know or amazon or whoever in monetizing all of the data and keeping you know the dark patterns you know their user agreements and things so that it's very difficult to detect you know what's actually being shared with whom and that i think is that that's got to change but it's yeah but
Renee:What do you think of the useful use of that kind of stuff. So the useful use of those metaglasses, like why don't Boeing engineers wear those when they're building planes?
Marc:Right?
Renee:And AI is good enough to know like, oh, of all the video I saw as the AI machine saw, there's a couple places where I think you guys need to go double check because it didn't follow what it was supposed to follow. Like there are, I think there are useful cases of that kind of stuff. I just don't think it's walking around, you know, on a street, using it to figure out who that girl is. Like, I just don't think it's that. Right. All right. So anyway, the Coke machine just wanted to tell you if the soda was cold. That was the entire ambition. No manifesto about cyber physical systems. No grand vision of smart cities or autonomous grids. Just a hallway, some lazy grad students, and a desire to avoid warm disappointment. But that small act, giving a physical object the ability to report its internal state, quietly shifted the trajectory of computing. From here, we layered identity onto objects, memory into the cloud, sensors into infrastructures, and models into the environment. Now our thermostat anticipates your arrival before you turn the key. Your watch tracks your stress patterns across months and nudges you toward regulation. Your car updates itself overnight, rewriting parts of its own operating system while parked in the driveway. That began, what would begin at a vending machine answering a question became a world that answers everything before we even talk.
Marc:The physical world has a feedback loop now, right? Sensors collect signals, models interpret them, actuators adjust conditions, and that cycle repeats continuously at planetary scale now. Traffic networks reroute dynamically, energy grids rebalance in real time. Buildings anticipate occupancy. Systems sense at a grand scale. Systems respond. Systems optimize. And optimization always moves toward an objective function. Efficiency, throughput, stability, comfort. The shift from reporting to closed-loop autonomy happened gradually. Most people experience it as a convenience.
Renee:Oh my gosh. And once the world has a feedback loop, it doesn't just observe. It adjusts. It anticipates. It optimizes. Sometimes it optimizes temperature or energy consumption. Sometimes it optimizes traffic flow. And sometimes, even more subtly, it optimizes behavior. The environment begins to shape us as much as we shape it. We started with a soda machine that prevented a minor inconvenience. We ended with infrastructure that quietly learns our patterns and adapts around them. The hallway gained metadata. The city gained telemetry. The planet gained instrumentation.
Marc:So 44 years on from a vending machine to planetary scale, cyber physical systems. And most of it happened slowly enough that each step felt like a minor upgrade.
Renee:We shouldn't fall for that. It's the boiling frog. Like, why do we fall for this? Like, why are we that lame? So anyway, you guys, don't be lame. Thanks for nerding out with us. If you enjoyed this deep dive into how matter became data and data became decision making infrastructure, Share this episode, leave a review, and keep asking uncomfortable questions about the systems humming quietly around you.
Marc:Next time your car updates itself at 3am, or your thermostat adjusts before you walk in the door, that's a closed loop system making decisions about your environment, quietly, without asking you, Renee.
Renee:I need a drink.
Marc:Thanks, everybody.
Renee:Thanks, Marc.
