Not so dumb.. cost point is the question.
Like any electrical device, there is lots of diagnostic information in the controller, that can't easily be communicated, and lots more that would be useful to add if only the controller could communicate with the owner.. all without sniffing the milk or replacing the paper shopping list with an expensive tablet:
o defrost cycles
o diagnostic info on the pump, compressor, water-filter (if it has one), host-zones
o raising alarm if door not properly closed/blocked
o decay chemical detection for rotting food for cleaning cycles
o smart boost: don't take a Phd to work out that loading 10kg of stuff into a fridge will need a boost
Granted, an IoT fridge is a lot less useful than an IoT boiler (carbon monoxide), but the benefits accrue when you combine simple sensors with electronic shopping (beer does not generally come in four litre bottles, but milk does) - think scales in self checkout.
A fridge that pings you to get more beer because you're running low on a world-cup night will easily outsell 3DTV.
Biting the hand that feeds IT
