Why robots and smart technology aren’t revolutionizing your house

Software is eating the world, wrote venture capitalist Marc Andreessen in a widely read 2011 article for the Wall Street Journal. And in many industries he's right. Software has had a huge impact in industries that involve manipulating, storing, and transmitting information, like photography, music, and travel reservations.

Yet progress has been slower when it comes to manipulating the physical world. Since The Jetsons introduced Rosie the Robot half a century ago, people have dreamed of smart robots saving us from the drudgery of housework. But companies trying to make this kind of vision a reality have been struggling.

Last month I wrote about Nest, whose popular smart thermostat made it a poster child for smart homes. But the company, which was acquired by Google in 2014, has struggled to develop new products, raising questions about whether Google overpaid for the company.

A similar story can be told about iRobot, the company behind the Roomba robotic vacuum cleaner. The company is hardly a failure, having sold 15 million units since it was introduced in 2002. But the Roomba remains a niche product, and iRobot hasn't come up with another hit.

These companies are struggling for similar reasons: Their products demand too much from their users while providing too little value in return.

Getting a Roomba won't change your life

Last year, iRobot sold 2.4 million Roombas. By any reasonable metric, that's a successful product. But in a nation of 320 million people (not to mention a world with more than 7 billion people), it's still a niche product. The vast majority of American households don't have a Roomba or any other robot vacuum cleaner and seem to be in no hurry to buy one.

And if you talk to Roomba owners, it's not hard to see why. "It gets stuck a lot," my Vox colleague Sarah Kliff told me. "I can't really leave it at home unsupervised."

Sarah has a table with a curved metal bottom that her Roomba finds fiendishly difficult to navigate. Often she'll come home to find that it drove up the table's leg and got stranded, the cleaning job unfinished. The Roomba also terrifies Sarah's dog.

I have a Roomba, and it has also had problems getting stuck. But I've also found that it just doesn't save me that much time. I still have to tidy up the room before letting the Roomba loose. Then when it's done, I have to empty the dustbin and — often — dig out debris that got caught in the rollers. It's not as much work as using an old-fashioned vacuum cleaner, but it's not that much less work.

And then there are the times when the Roomba wreaks havoc. "My old roommate had a Roomba that ran into my mirror," Sarah told me. "The mirror toppled over and broke."

One day, my Roomba got ahold of a spool of thread. When I got home, it had unwound the entire spool and wrapped it around the cleaning brush roll. It took several minutes to get it unwound, and I had to throw away the rat's nest of thread that was left.

I have a $399 Roomba 650, and iRobot just introduced a new high-end model, the $899 Roomba 980, which comes with a built-in camera, a longer-lasting battery, and other improvements. But as Fortune's Kif Leswing pointed out in a review last October, these improvements only get you so far. The longer battery life doesn't help if the dustbin gets full or your home has multiple levels. And the latest Roomba seems about as clumsy as its cheaper cousins — Leswing says it "beached itself on the legs of my Ikea Poang chair." And it ate one of his cat toys, damaging one of the robot's wheels.

Why it's hard for smart appliances to add value

The Roomba is by far the iRobot's most successful product. Over the years, the company has built a couple of mopping robots, a pool-cleaning robot, and a device for cleaning out your gutters. None of them have been big hits.

Other companies have tried to create internet-connected lawn sprinklers, crock pots, and lightbulbs.

A fundamental problem here is that for many tasks in the physical world, there just isn't that much room for software and complex electronics to add value.

The home appliances that have done the most to improve people's lives are the ones like dishwashers and washing machines that took a really time-consuming and tedious task and made it dramatically faster.

But in many cases, the pre-internet devices in our homes are already pretty good. There isn't a ton of room for further improvement. People don't spend a lot of time adjusting their thermostats, so the better interface on a Nest Learning Thermostat doesn't add a ton of value. Smart lightbulbs or robotic gutter cleaners seem even more a solution in search of a problem.

Machines add the most value when they can be operated at scale in a controlled environment — washing machines and dishwashers are useful because you can wash dozens of dishes or shirts at the same time. And because all the action happens inside the machine, there's less room for unpleasant surprises — like a stray cat toy getting into the gears.

In contrast, home robots and connected home devices are trying to operate in the chaotic and nonstandardized environment of a modern home. It's an inherently more difficult problem to design a product that will work flawlessly in a wide variety of different home types.

And this is a reason to be skeptical that we'll see rapid progress in household robotics or smart homes in the coming years. It has proven difficult to build a robot vacuum cleaner or a smart thermostat that's a big hit with the public. And other home automation tasks — like iRobot's mopping robots — have been even less popular than that. The concept of smart homes and cleaning robots sounds appealing in theory, but making it useful in practice is surprisingly difficult.