Sylvia Else wrote:

[attributions elided : ]

There's not much point in having a comfort feature if it's likely to
get turned off at the time of greatest need.

It's not "turned off", per se. Rather, it is *deferred*.
I.e., maybe 10 minutes later your ACbrrr will kick in
instead of at the (slightly) earlier time when the
thermostat called for cooling.

Chances are, a normal user wouldn't really perceive the fact
that the ACbrrr kicked in "late" -- the house may rise some
fraction of a degree beyond the thermostat's set point?
But, the *idea* of this is unnerving to many people:
"Oh, I am going to be *so* uncomfortable!"

OTOH, if they were to raise their thermostat's setpoint
that degree (or two?) all the time, they would probably not
notice the difference.

But, as I've commented elsewhere, as soon as the AC is powered up it
will stay on for longer to bring the temperature down again.

Yes, but not *much* longer. All you are doing is shifting (in time)
when the house will be "as comfortable" again. E.g., if you
had opened the front door (to carry a large piece of furniture
in/out) you would similarly have disrupted the comfort level
in the house for a short period of time. (if you've ever
"moved" in the summer time, you'd understand : ) But, that
comfort level is once again restored, ultimately -- *later*.

The big (psychological) problem with load shedding is that
folks don't feel like they have any control over it so
they assume it will be uncomfortable, etc.

OTOH, if the technology was implemented as an *auction*
(I am merely trying to make a point) in which the utility
could alert participants:

"Hi, we need folks to shed some load. We are currently offering
$X for you to shed Y load for Z minutes"

and then respond to folks who have accepted this offer by
adjusting their *new* "bid price" (up or down):

"Due to the overwhelming acceptance of our previous offer, we
are now only willing to offer $x (i.e., x X) for you to shed
Y load for Z minutes"

or:

"Hmmm... we haven't had many takers of are offer at $X so we are
now sweetiening our offer to $XX for Y load for Z minutes"

Then I suspect most folks would probably complain that watching
for these "offers" is too tedious: "Can't you give me a
SWITCH that I can throw that says "I am willing to accept
ALL offers of $X or more?" (which, in effect, is what the
tariffs do -- except you have to make this commitment up
front!

There is an energy saving, but it's very much a second order effect,
arising form the slightly higher average temperature.

The goal isn't to save energy (though I think the laws of thermo
say you *do* save in this case). Rather, the goal is to
get you to *shift* your energy consumption (in time).

I'd only go with it if the saving were enough to finance a petrol
generator to supply the power instead.

frown Defeats the purpose. The utility can produce electricity
far cheaper/cleaner than you can.

I'm not so sure that's necessarily true for extreme peak loads. They not
only have to generate the power, which is likely to be done using diesel
or some such, but they also have to deliver it to me, which involves
transmission infrastructure which is only used during these extreme peak
loads, but which is there all the time.

The infrastructure is sized for some percentage above nominal.
Of course, going too far *beyond* that causes things to *break*
(hence the blackouts that become newsworthy).

But, things like diesel/coal/gas fired plants that are there
*deliberately* to respond to these short term fluctuations in
demand could, theoretically, be eliminiated if the demand could
be "leveled".

Nuclear power plants, for example, like to put out a steady amount
of power (can we *please* not let this discussion digress into
the pros and cons of nuclear power? : ) which doesn't lend
itself to rapid response. If you had fill-in-the-blank power
source that was ecnomical to operate and had this characteristic,
then you would have a big incentive to coerce users into
adapting their usage patterns to match.

Also, it is important to define what crteria you actually want to
optimize. E.g. "efficiency" can be defined in a lot of different
ways -- many of which are inconsistent with each other :-/