The idea seems to be that Imp will reduce the development / maintenance costs for manufacturers who would want to put these kinds of features into their products, making it more likely that this "internet of things" ecosystem evolves than it would be if the manufacturers themselves were trying to do it with their own costly, buggy, non-interoperable technology stacks.

This "sell to the manufacturers, who will in turn sell to the customers" bank-shot approach will have to overcome the limitations / drawbacks sn00p mentions above, which all seem like legitimate and reasonable complaints to me. They really have to get it right out of the gate, with some big-name manufacturers and some seriously useful features. I don't think they can sell this as a home tinkerer type thing like the Arduino, because tinkerers aren't going to want to deal with a somewhat-expensive-for-what-it-can-do-on-day-one cloud service.
_________________________
- Tony C
my empeg stuff