The internet of things won't be anything unless someone can figure out a better user story than "turn your house lights on and off from anywhere!"
But the implications (especially for security) are huge. Not just because of new bugs or design flaws but also because the momentum (and media frenzy) it creates puts these old (buggy) technologies into the limelight of security researchers. E.g. stuff that was lurking in old protocols suddenly becomes relevant (want to stop a train? https://media.ccc.de/v/32c3-7490-the_great_train_cyber_robbe... ).
There are countless applications not just for BAC but also for logistics, power-grids, industrial automation, e-health, insurance, ... where the IoT is already actively pumping out use-cases, product ideas and actual products.
I don't know if it currently works like this in Japan, but the idea is that with a reliable enough warning system, vital infrastructure could automatically respond to the few seconds warning an earthquake alert might provide.
Hearing that, an Internet of Things suddenly seemed to me a little more serious and useful. Of course that doesn't excuse poor security or sloppy design. It only makes it that much more important.
The high speed rail system has its own detectors, both along the rail lines and at the coast.
And then when I see hypothetical discussions about smart appliances ("a fridge that can automatically order food when you're low!", "a dishwasher that can tweet when the dishes are done"), the thing that I really want are open-source hackable everyday appliances, not feature-rich ones.
E.g., I want to be able to confidently replace a busted controller in my dishwasher without necessarily having to shell out $100 to maytag for one that was only produced for a single year.
Sadly, yes, this is a wish born out of experience.
Not that I see a need for something like this, but existence of such device could deter a bit companies from shoving such things to every appliance.
Part of the problem with IoT, as others have pointed out, is lack of a compelling use case for most of the products. I don't think this is a fatal flaw unless all of the also-rans and wannabes and opportunists suck all the air out of the market before someone with a compelling application (and decent reliability!) can gain traction in a non-privacy-invasive way.
It is also a bit like uncanny valley. Because if we will at some point build general purpose robots they will have microphones and cameras. On path to this there is a point at which you must trust having microphone and cameras from less useful devices pointed at you. If not how can we have off-the-shelf general purpose robots at our disposal? It also creeps me out... How can we achieve this? Some can say we shouldn't.
Try and turn it around.
Instead of calling it Internet of things which indicate it it's own thing, instead think of it as things connected to the internet.
Now you realize that this is already happening and it's a combination of our phones and our fridges, watering systems and load balancers all able to communicate.
So the user stories are already there we just don't think about them as that because of this claim of a separate internet. It's not. It's the same.
In a way, it's the problem of the current Internet, but exported to the hardware world. Just like NATs killed publishing, they also kill sanity in the design of connected things. I wish we could deploy IPv6 faster. Yes, I want my fridge to have its own IP - so that I won't have to proxy the communication through some random fly-by-night web startup.
He loves it.
PS: He's also got the IoT lights :P