The Internet of (Poorly Working) Things

In the mythical Land of Theory, where everything “just works,” we can connect all the objects in our lives. We have the sensors, the wireless networks, and the computing power, but progress is slow if not comically wrong. Why?

It was 20 years ago this month the cover of the adorably geeky Boardwatch Magazine, a journal dedicated to the world of Bulletin Board Systems (BBS, remember them?), featured bearded professor Vint Cerf, the grandfather of the internet.

Cerf’s pithy “IP On Everything” prophecy was simple, resonant and inexorable: All objects in our lives will someday feature an IP stack.

Two decades later, where are we? Moore’s Law has given us a 2¹⁰ to 2¹³ improvement in chip performance—that’s 1,000 to 8,000 times more computing power. Personal computers and smartphones are everywhere, well adopted and put to productive and enjoyable uses.

(As an aside, smartphones are actually the most rapidly and broadly adopted product of all time, leading observers to conjecture we’ll never see anything like it ever again. False prophets and worried CEOs looking for the next big thing, take note.)

With a three-orders-of-magnitude power increase and oceans of wireless data packets, surely we can endow our everyday objects with all sorts of sensory and connectivity magic.

So, how is it that we don’t have connected objects that Just Work?

Before we offer an answer to this question, I beseech you to follow the Internet of Shit (@internetofshit) on Twitter. Trust me, you won’t regret it:

With more than 2,000 tweets and 130,000 followers, IoS is an eye-opening poke at today’s IoT (internet of things). To whet your appetite, regard the very smart toaster:

Or consider the “famous” $700 connected juicer:

Anything you can imagine—plus a few you don’t want to.

Of course, not all connected devices are so easily mocked; some devices are dead serious: home security, HVAC, almost any kitchen appliance—even our very smart toaster. And it’s not that the IoT doesn’t work. The situation is actually worse than that: The IoT randomly works. Devices stop and restart, and require visits to unsupportive customer support pages and helpless “Your call is important to us” help lines (and now we have chatbots). If you think I exaggerate, google “Nest trouble” or “smart bulbs trouble.”

We don’t have to look around much to find the culprit: with its razor-thin margins, the Consumer Electronics culture offers a big fat target for our inquisition.

With a tight budget and limited software know-how, Consumer Electronics product development teams are issued orders from on high: Get on the IoT train…now! They buy the cheapest possible processor, grab some software from the open source shelves, throw on a skimpy user interface, hastily assemble and test, and ship it.

The meager budget doesn’t leave much room for user instructions and customer support, and it sometimes leads to dubious design solutions. I once served on the board of a company that contemplated acquiring a home networking business. In our due-diligence work, we found many of their software modules were simply lifted from another company. Not open source—pure misappropriation. I doubt this was an isolated case.

After a device has made it to the market, the real fun begins. Software updates are a problem when your connected lock or your connected car is in the middle of an update and you need to get into the house or drive to the emergency room.

There’s another reason for disappointing adoption: Complexity.

It’s one thing to power cycle your router when your internet connection slows to a halt. How do you debug a network of 10, 20, or more connected objects in your house—from light bulbs to locks and sprinklers—that run on a mixture of Wi-Fi and something else, such as ZigBee for LED lights that require a special bridge?

As expressed in Barron’s “The Internet of Things Is Too Confusing,” [paywall]:

Even tech experts are frustrated by the complexity. “In one out of 10 cases, I get in my car and my smartphone just won’t connect to the stereo system,” says Cees Links, a pioneer of WiFi wireless networking. “And in those instances when it doesn’t work, there’s no clear reason why.”

Then, there’s the truly ugly side of consumer IoT: security, or the lack of it.

The lackadaisical approach—to be polite—to software leaves many connections open to hackers who can see passwords exchanged in clear text on home Wi-Fi while they sit in a car parked outside a house. Or we see 100 million Volkswagen cars are open to wireless hacking. Using the One Cockroach Theory, how many more other makes of cars will be found to be insecure?

Speaking of cars, Apple’s CarPlay sheds more light on connected devices. After a long wait (it was announced in 2010), CarPlay is now available on “selected” models from Ford, Honda, Fiat Chrysler, GM and others.

According to the initial reports, it’s good, not great. Of course, CarPlay is a software layer on someone else’s hardware/software infotainment system. In other words, it’s a situation where Apple doesn’t control the whole hardware/software stack, similar to the relationship between Windows and PC makers, or Android and handset manufacturers.

A look at the so-called Smart TV reinforces the observation about CE culture. In less than two years, the CPU inside the TV quickly becomes obsolete and can’t be upgraded, while the display itself easily lasts a decade and software updates are perfunctory, if they happen at all.

Word of mouth is still the most potent marketing weapon, especially when a product doesn’t work. Two years ago, Apple introduced HomeKit, “a framework for communicating with and controlling connected accessories in a user’s home.”

Now, look at how many HomeKit devices are on the shelves at Apple Stores—pardon—venues. The number isn’t growing. Some make a brief appearance and are never to be seen again after too many returns or customer service calls.

On the bright side, we do have an internet of things that works: The industrial version. Modern buildings are equipped with sensors, connected HVAC, security, and power management. But there’s no scrounging on the cost of devices, they must work and last, and the building owner has a technical team to install, maintain and run the whole system.

This is the lesson Consumer Electronics makers must learn: A successful IoT can’t be built on the cheap. And this is precisely what we see in the Amazon Echo.

Echo is a connected Wi-Fi/Bluetooth speaker and microphone that features “Alexa,” an agent that responds to your queries, plays music, controls devices and even starts your car (if it’s the right make and model).

Amazon spared no expense in developing and supporting the Echo, and the investment has been repaid by excellent word-of-mouth. This is the antithesis of the cheap CE approach—it's a well-funded company with proven technical expertise in Cloud services, a successful history with Kindle devices, and, above all, a determined group playing the long game with Jeff Bezos at the helm. (Let’s add in the “helpful failure” of the Fire smartphone.)

Speaking of support, Echo customers—and I have one sitting in my office—get a weekly newsletter that announces new features and suggests new uses. I know I have this fixation of weekly writings, but how many companies write to users of one of their products every week?

I have little doubt the success of Amazon’s Echo inspired Google Home—still a demo at this stage. I hope to see others.