The Marketplace of Things

The Internet of Things (IoT) is a classic Big Idea, a techno-animist dream of the collective potential of all of the wearable technologies, smart houses, and wired industries^[1]. As wireless-enabled devices have proliferated in homes and industry, a massive flow and interchange of data has emerged, and with it the potential to transform all of the “dead” objects in our world into a dynamic connected whole.

The digital IoT (aka the I) has had a rather big run the last couple of decades. Over that time the commercial heart of the internet has moved from the content we browse to all of the myriad data we generate transacting with Facebook, Google, Amazon, and every other website. Of course, it’s even more complex because it’s not just the websites we’re visiting that are collecting information on us, but all of the lurkers following us throughout our digital day^[2]. Websites and apps are highly engineered and instrumented machines; our every action in them creates a spray of data from which they can infer a progressively richer and richer model of us^[3]. Many companies trade this data for helpful insights (I’m quite fond of Google’s reminders to leave now or be late for my next meeting^[4]). Others feel like a stalkery ex-boyfriend destined to violate a court-order. In the end, the prolific data collection on the internet grew out of the collaboration of engineers and MBAs moving faster than any existing social institution could keep pace, promising to optimize the digital world for all its inhabitants. They laid down norms and standards of data ownership before anyone, even themselves, really understood what was at stake.

Just as the World Wide Web can be instrumented and optimized, so can our wide world. For most of us, the first true expression of the internet of things was the smartphone. Packed with gyroscopes, accelerometers, GPSs, light sensors, microphones, cameras, and so much more, every phone turns us all into walking laboratories. And with all of that explicit data–contacts, emails, texts, apps, songs, videos, and more–your phone would make any FBI dossier feel jealous and pathetic by comparison. It’s not just that mass of data, though, that makes your phone the cornerstone of the consumer IoT; it’s the connectedness. All of that data is constantly being shared out to distant company servers and back to the other local devices that are expanding the internet of things beyond the smartphone.

As I walk in the door at the end of the day, my first words are “Okay Google, turn on all the lights.” That command passes through my Google mesh wifi network to the Philips Hue hub, which then commands various smart light bulbs placed around the house to come alive in selected colors. My phone and tablet are updated with the state of the lights and network. They in turn automatically update Google, Amazon, and others on my location. Phone and fitness tracker, three inches apart from one another, exchange information via distant platform hubs about my activity levels and perhaps even my emotions, which can then further refine the colors and intensity of my lights to match my mood.

Preprogrammed schedules turn my children’s wifi access on and off so that they don’t spend hours watching videos while I prep dinner^[5]. Smart sensors connect my house to the power grid, helping to optimize power utilization through the entire region. Many others own connected refrigerators, cameras, cars, bikes, crock pots, and toilets^[6]. With our smart phones at the nominal center of it all, the promise of the IoT home is the same as the modern internet: “The more data we collect about you and the more objects we connect inside your home, the more we can optimize your life.” It’s an alluring promise with real substance. Add in smart clothing and medicine and transportation networks–every single component connected together–and the consumer IoT is a massive effort to turn our physical world into a new web, not to mention a huge new market full of products to sell.

As sexy as the IoT smart home might be, for many experts and companies the consumer IoT is not the most exciting market. Industry, freed from many of the ethical and marketing challenges of consumer markets, has become the proving ground of IoT. Mundane, even old-fashioned, technologies like windmills are transformed when every blade is brought online as its own robot, packed with sensors and actuators. Efficiency of the entire system is optimized by creating closed loop systems converting the streams of data produced by every component into minute changes in blade angle, torsional loads, and numerous other degrees of freedom. Entire buildings, production lines, distribution channels, and fleets of heavy equipment come alive in the industrial IoT. The promise again is optimization: squeeze out all of the inefficiencies in your business and maximize the productivity.

The promise is real, excitingly so, particularly in industrial IoT. When a single company engineers the entire system (e.g., a jet engine, a train, a power plant) the platforms can create truly astounding gains. But there’s one major flaw at the heart of IoT. It is a platform play, a walled garden. Every major tech company, and many others chasing after them, are trying to own the internet of things by becoming the platform on which it runs. Most of those smart objects in our homes and businesses aren’t talking to each other; they’re passing messages back and forth via centralized platforms without a common language (e.g. standard internet protocols) with which to communicate. The possibilities of intelligent homes, businesses, and even communities are real but submerged in the interests of the platform owners. Because it’s not truly open and distributed, IoT can never grow into what we can imagine and will always live under the yoke of who owns our data.

This problem isn’t unique to IoT. In some ways, Facebook, Google, and others know me better than I know myself. They build models of me–digital ghosts that will live on long after I’m gone. The question of who should own it all has received increasing scrutiny recently, with some even proposing anti-trust legislation against the biggest data aggregators. As it stands now, though, for all practical purposes those companies own the data I generate and the models of me grown from it. So, because of the rather peculiar wild west history of the internet, Facebook owns my ghost, my avatar, my other self, the distributed me.

Similarly, IoT has become a competition between dominant platforms, with control and communication passing through centralized hubs governed by these dominant businesses. Data from wearables, medical devices, phones, homes, buildings, and heavy industry, streams to these platforms, leaving little room for the IoT enabled devices to directly (and privately) talk to each other. It needn’t be that way.

Of course, this is the vision of Blockchain, the distributed, decentralized cryptographic algorithm behind Bitcoin and so many other techno-anarchist wet dreams. I’ve never seen a solution more desperately in need of an actual problem to solve. Could you imagine if my light bulb needed to wait for a massive quasi-prime to get factorized just so I could find my keys? But the idea of massive, distributed trust networks allowing data transactions without a centralized hub is a rich and potent area of research that pre-dates Bitcoin. This research could enable a dramatically different vision of IoT, a Marketplace of Things (MoT^[7]).

In a marketplace of things, devices don’t network with a proprietary server, they directly talk to each other. They share information and capability in an open market using distributed trust technologies to support their transactions. My lights want to know my mood to better set their hue and luminance, and so they offer facial expression information (previously exchanged with embedded home cameras in exchange for movement pattern data collected across its meshed bulb network) to my phone in exchange for affective computing data it’s been collecting throughout the day. My fridge exchanges updated food information with a health and wellness subnetwork distributed across my clothing, phone, and smart shoes, which have collectivized their negotiations to improve their market power against product marketing campaigns negotiating for my gustatory attention. Every device transacts directly with every other device, with information flowing in complex patterns throughout my local home network. None of this has any need for a distant corporate server capturing the processes and taking a percentage (often 100%) of every transaction. It is complicated and messy and daunting^[8] and truly needs to happen. The centralized IoT needs to be disrupted if we ever hope to turn its walled garden^[9] into the complex, evolving ecosystem it could be.

For me, the most exciting thing about MoT isn’t just some Libertarian ideal. It’s more than freedom from centralized data ownership by an increasingly small number of very big companies. It is the ability of such a system to evolve in ways no planned economy can. Locked down in a platform, IoT will only ever be what its small number of managers can envision. The near-universal fear of disruptive innovation will suppress its potential, just as monopolies sap the efficiency out of otherwise open marketplaces. And, of course, there is one stakeholder conspicuously missing from data transactions on nearly all internet and IoT platforms: the user. How can there ever truly be an open market if the central figure in every transaction can only participate by giving total proxy to the platform owner by virtue of an End User Licensing Agreement? We get no direct say in the ads we see, the data we share, or the use of that data beyond an intentionally obfuscated all-or-nothing opt-in. In an open market we can inject our own proxy into the negotiations: you need to pay me directly to show that ad; I’d see that one for free; I’d rather “pay” for access to this site than experience any of those ads. This means that I own my data and the digital ghost it generates. It’s that ghost which becomes my avatar in those negotiations, and is negotiating for my interests. I don’t have to think ill of Google, Facebook, Tencent, or any other company to understand that they can never truly represent my needs above their own.

Even all of the market mysticism I’m shoveling above isn’t why I find MoT exciting. In this marketplace, every device is an independent processing unit networked into a distributed whole. Every smart bulb, phone, juicer, fridge, car (and all of its subcomponents), thermostat, camera, and more draw in sensor data about their environment, combine it with other information processes from other “things”, integrate the new knowledge, update internal states, and select actions (policies). These processing units offer their new knowledge and states out to the market for other units to consume and add meaning to their own processing.

As transactions accumulate and trust increases in the market, networks of relationships between units in the market strengthen (or weaken for freeloaders and exploiters). Signals become data, becoming information, and then knowledge and actions, with the entire market network learning and developing over time...that begins to sound a great deal like a brain.

What excites me about an open marketplace of things is its potential to become a true distributed cognitive system. Rather than eyes, ears, noses, and other sensors attached to a fixed (if slightly jiggly) body exploring the world, this cognitive system is an ever-changing swarm of networked devices communicating in ad hoc circuits. Individually they pursue their core objectives (e.g., set the right light levels, read the identity of the faces in the room, suggest the right song, etc.) just as a simple organism serves simple objectives (e.g., eat, mate, avoid being eaten, etc.). Collectively they might begin to abstract up from their low-level needs, learning and exploring the world^[10]. Every device in the MoT is a sense and an action. Every one is an organ to explore and exploit its environment. The learning is both integrated across local circuits in the network and traded off to distant regions. Complex concepts emerge in the dynamics of meaning and action flowing across markets, enhancing every individual unit’s efficiency. Artificial ecosystems like this tend toward relatively simplistic dynamics and lack diverse speciation. In MoT, diversity is maintained by the simple fact that every new “thing” has new information of some marginal value. These differences in sensors and processing improve the information sampling of the cognitive whole^[11].

Its cognition changes as people and their devices walk in and out of it. You change the collective just by arriving at work^[12].

This is all wild and crazy speculation, but if I were looking for the emergence of AGI, true human-scale intelligence, barring a massive leap in AI research, it would be in that wild marketplace. The platform business built around IoT will provide unconscionably large returns to those companies that control and govern the interactions between all of the emerging smart devices in our world. It will be simple and efficient and highly predictable. It will never dream about the people scurrying around inside it.

An open marketplace of things, an entire ecology of interacting AIs, will be messy and complex and surprising. The potential winners and losers releasing products and services into the marketplace will be unpredictable and ever-changing^[13]. In those markets of wild ecosystems of distributed thinking things, perhaps something truly amazing will happen. Perhaps the building wakes up.