Smart meters in the UK will come with an option to shut things off in households, which is unsettling for many people given the state of cyber-security we have witnessed recently:

http://www.theguardian.com/technology/2013/aug/18/smart-meters-uk-hacking-electricity

I'm glad this is being said in an important newspaper. Actually, most of the critique towards smart meters in recent years has been about whether we want to give up usage data, given possible privacy violations. That is a discussion worth having, but in my opinion the discussion about the remote off-switch switch is much more important. I never saw the need for it and during the few years that I've been interested in this field, I think I've seen many who have previously supported it come to the same conclusion. So maybe that the off-switch is still in the specification is actually a state of several years ago and doesn't actually have overwhelming support among experts anymore - you know there is always terrible inertia in such standards.

Actually, support of the off-switch was never so that you could unhook non-paying customers, as the article says, but rather a dream of a novel tool to optimise the grid balance. I believe it comes down to central versus decentral optimisation/planning. If the utility or anyone who contracted flexibility from households can then (within this contract) act on their behalf (shutting something off temporarily), they now need to solve this optimization problem all alone, at their central point: when to regulate which connections how much, given outside circumstances. If only a market price is communicated, the problem becomes decentral: each connection makes their own decision, based on the price.

There are a couple points to make between these two (roughly-painted) approaches, I won't be able to make them all here. Let me just make two:

1. For the system designer, the central planning is a rather classic optimization problem while the decentral implementation one is rather a market design problem, which lets control signals (prices) develop (within some well-defined bounds) by the decisions of the many and it also leaves open a space for technological innovation at household level. A classic optimization problem is more appealing because we feel we have the right hammer for that nail available. However, we will then need to make strong assumptions about the underlying model. This also holds for market design, but at least there are less points of failure and in general less hybris on the part of the designers (especially in a world that luckily is as tightly regulated as the electricity world is).
2. Such adjustments for grid balancing do not make sense for small appliances. There are only a couple use cases which provide large enough room for flexibility, like industrial freezers. On the household level, which are relevant for this smart meter discussion, we have not many yet. We can expect to have electric cars and heat pumps, though. Those two use cases will be significant in size, but also in importance. Who would like to give up all control about transportation and temperature control to some utility with an off-switch?