IoT Is Not Mobile on the Wall

I’ve started listening to a16z podcasts during some meals. I recommend them if you're interested in the future of technology. About a month ago I listened to this fascinating conversation between Benedict Evans, Preethi Kasireddy, and Zal Bilmoria, though I'm just getting around to writing about it now (better late than never). Post-CES, they try to tackle the question of, where is this Internet of Things we keep hearing about? Is there an Internet of Things, or is it just “things connected to the Internet”?

Around 1:49 Benedict Evans makes the most compelling argument: our grandparents could have told us how many electric motors they owned: one in the car, one in the fridge, and one in the vacuum cleaner; and now motors are so common that the side mirror of your car probably has a dozen. But no one goes out and buys electric motors; they buy a microwave, a blender, a coffee machine, etc. – devices that solve a problem that just happen have an electric motor as part of the solution. So it is with the IoT: other than early adopters. it will enter our lives slowly, as we buy device that solve our problems by adding a bit more intelligence to existing devices, a bit at a time.

The analogy makes sense, but it has one fatal flaw: the functionality of electric motors is self-contained (with the exception of transportation, but we are discussing household devices here), whereas connected devices are not self-contained by definition.
When I use my blender, the result occurs wholly within the blender; it doesn’t interface with the coffee machine, and it doesn't generate externalities that I have to worry about, whether it is a privacy concern or merely the cognitive load of making the best use of my data.

Connected devices don’t just create a result, they create data. And that requires human beings to interpret and act on that data, and that’s where it becomes a conundrum for the market. Because if there’s one thing we don’t need more of, it’s data. Beyond early adopters and programmers, for whom this sort of thing is second nature, no one wants the cognitive load of having to do data analytics or write if this then that kind of rules to get the value of a device they just bought. So, we have to create devices and systems that are self-contained, or at least provide the user with the illusion of self-containment. Interoperability won’t cut it; it has to “just work.”

This will require a tremendous amount of effort on the part of OEMs and vendors towards simplification, but I don’t buy that that means that smartphone apps will be the enabler. Not because I have any reason to doubt the smartphone - and no doubt that in the short term apps will be the one of the primary centralized interfaces - but because it’s just too easy an answer to throw out to a novel problem. It’s like the metaphor that to the man with a hammer, everything is a nail; we have a smartphone, so everything is an app. I think we’re too early in the life cycle of connected devices to be able to conceptualize the ultimate interface; we lack sufficient user and market experience.

To offer my own rough analogy, the Internet of Things is where mobile was before the iPhone. The components were all there, but it was missing the unifying paradigm. Knowing everything was there didn’t matter. Remember pre-iPhone, when manufacturers were trying to force WIMP-based interfaces onto our tiny phones with their tiny screens, and mobile application discovery and installation was limited to only the most tech savvy among us? Then Apple comes out with the MPG UI paradigm and the app store, and suddenly everything just works, and it was obvious that this is how mobile computing should be done (obvious at least to the early adopters; remember people swearing there was no way they could use a phone without a keyboard?). The best way to predict the future of IoT too will be to invent it.

So what are the possibilities, in the meantime? Industrial applications, which Zal mentions in the podcast, seems like one way to go. If using connected devices requires significant cognitive load, then it makes sense to deploy it initially in the workplace, where specialists are already doing the heavy cognitive lifting, systems planning, and data analysis.

Another possibility I see it is a market for IoT home contractors (it’s already beginning for builders). I don’t have to know how to wire my home, or what electrical supplies to buy for all my rooms to have electrical outlets, cable jack, and fuses; I just hire an electrician. Why shouldn’t I have the same option for connected devices? The electrician could even set me up with custom dashboards and instructions for my personalized killer apps, and offer ongoing tech support, either on an as-needed basis or as a subscription (why is this starting to sound like cable?)

One potential issue is that it's still too early in the game to agree on what a “standard” connected home setup should include. I think this still work on high end of the market, as a concierge product, and then gradually moves downmarket as prices come down and use case standardization evolves. Companies like Wink and Nest might even do well to offer this service, both as a way to create sell-through, but also to keep a pulse on where consumers’ heads are at with as the market evolves. The customers that this would attract might even be pragmatists rather than early adopters, providing a beachhead into the mainstream part of the proverbial chasm.