There is a very good post by Adam Dunkels about the difficulties with IoT projects. His analysis is excellent, but he does not stand back far enough. People do not know what IoT is.
“IoT” has almost no stable folk definition. They get the fact that it is wireless, meaning invisible. “Somehow” invisible things can connect to visible things. After that it gets murky. Ask ten people what IoT means and you get fragments; “smart home stuff,” “the app on my e-bike,” “sensors,” “5G,” sometimes just a resigned shrug. There’s no shared mental model in the way there is for, say; “laptop,” “router” or “iPhone.”
Shared mental models
We probably need to unpack “shared mental model” before travelling further. Car, bike, dog, house,… terms with deep underpinnings across cultures. Your dog knows where your house is and guards it fiercely. Someone can know what a sensor is. People can use GPS location tools. They can connect to Wi-Fi. They can manipulate a database or a spreadsheet. They can show you an app. They can understand perfectly well that a machine somewhere sends information to somewhere else. And yet they can still have no mental object called “IoT.”
Humans are rather bad at intuitively perceiving systems whose important properties exist between (the connectivity layer or network of things) components rather than inside them. A car is comprehensible because the thing itself presents a coherent object: Car → engine → wheels → steering → movement. You can point at it. You can sit in it. You can crash it. You can fix it. The conceptual boundary is obvious and visible.
One of the cool quotes in Dunkel’s post is; …wireless is hard because it combines all the difficulties of the physical world with the complexities of software.” This is an elegant introduction to a snake pit.
Now take an industrial IoT installation. You’ve got: sensor → wiring/radio → gateway → network → protocol → cloud/server → database → application → human decision → physical intervention. Where, precisely, is the IoT? IoT is the functioning relationship between all of them.
The general public is being asked to comprehend something that doesn’t present itself as a discrete thing. That’s a hard cognitive proposition; Information originating in the physical world becomes actionable (if) somewhere else, (then) potentially causing something to happen back in the physical world.
The term was seeded
The crux is that industry has spent 15 years mansplaining everything from the component level upwards. The term was seeded, not a grassroots conception. “IoT” was coined analyst-side (Kevin Ashton, 1999, then heavily pushed by Cisco/IBM/Gartner-style market forecasting in the 2010s) to describe a market category they wanted to own, not a lived experience. Words like “app” or “selfie” win because they emerge from usage. People need to name the thing they are already doing; bmxers bmxing, farmers farming – iot ioting.
Public perception begins at the interface, with all that brilliantined glitter of a casino foyer one armed bandit. What is happening now is a quickening – we can initially define an object by what it is not, better than by what it is (qualification by subtraction). The industry is being forced to move from defining IoT by components to what a successful IoT system must have. That is a profound shift; fragmentation → consolidation → technological discipline → negative definition → positive definition.
Early IoT markets tolerate ridiculous complexity because everybody is experimenting. Nowadays somebody has to operate the thing for ten years. Not the maximum amount of connected technology you can deploy, but the minimum viable technical system capable of producing a sustained operational improvement. Questions are far less unrequited romantic longings, much more engineering and business requirements.
- Does it reduce operational workload?
- Does this integrate with the machinery I already own?
- Can I scale from 5 devices to 5000?
- Who owns the data?
- Can I move the data?
- If it breaks, can I fix it?
- Can my maintenance people support it?
- Can I change or upgrade connectivity options?
- Can I replace a sensor without replacing the platform?
- Can another supplier work on this system?
- Will this vendor be around in 5 years?
- What happens when my pilot funding ends?
The thing that may finally make IoT comprehensible is the gradual disappearance of IoT as a thing you have to think about. Much like electricity. Nobody needs to understand the electrical grid to “get” what a light switch does.
