Tyler Cowen reads my prediction about productivity growth in caring for old people, and says "I hereby take Megan to be a robot pessimist. "

For the relevant time period, yes. The boomers start retiring this year; by 75, they'll start having major life impairment. By 20 years hence, we will--if we are not to lose productivity--need a robot capable of feeding and bathing an old person without damaging them. I don't see that as particularly likely. The technical hurdles are tough, and so, it turns out, are the economics:

It's easy to walk through the Robotics Tech Zone at CES without ever realizing it. There are card tables strewn with a handful of gaudy brochures, booths that are completely empty, a handful of extroverted toys, and what appear, at first, to be many Roombas. Some of these roving disks zoom across smaller pens, one across a stage—all idly bouncing off their surroundings and mercifully unaware of how boring they are. That's because they're basically clones of the iPod of consumer robots: the Roomba.

In fact, these competitors don't really function any differently either; Yujin Robot's Plus A robotic vacuum, for instance, boasts a list of features that are identical to the latest Roombas, including pre-set cleaning times and the ability to automatically recharge its lithium-ion battery. With 2.5 million Roombas sold, and no one currently coming close to out-innovating its flagship model, iRobot has effectively zero competition. But that's not necessarily a good thing.

"One company doesn't make an industry," the company's CEO, Angle, has said for years. That's why iRobot developed Create, an open-source, mod-friendly version of Roomba that could not only empower the niche bot-hacking community, but serve as a research platform to help start-up robotics companies get their footing. As investors warm up to the concept of consumer robots, Angle claims it's becoming easier for newcomers to get access to capital. Still, he said, creating household bots is a nightmarish business proposition.

"Unlike with software, the margins are terrible," he said, citing 56 percent drop-off from software to robotics profits. "And you're building physical stuff. You have moving parts, gears operating in nasty environments. The robots are going to break." Initially, the Roomba was built to last 150 hours before failing, to meet European product standards. But considering how often the vacuum runs, that would have meant just six months of operation.

Even military robot-makers like Boeing and Foster-Miller, who are no strangers to engineering for endurance, would be crippled by the slim profit margins involved in consumer robotics, according to Angle. And while Samsung, General Electric and similar companies might have a better shot, he maintains that gaining a foothold is "ridiculously hard. It took us 10 years to develop the competence," Angle said, some of which included a toy deal with Hasbro that didn't pan out, but helped teach the company how to deal with production and quality control issues on a massive scale.