I understand that such power is used instanteously as soon as it is
generated.  But demand at a given moment must be constantly changing. 
So somebody turns off a light or shuts down a motor. What happened to
the power that would have kept that light or motor running had someone
not turned them off?  Can you stop generating suddenly-unneeded power
in, say, a nanosecond--or less?  I don't see how--even a nanosecond is not
instanteous  Or how does it work if suddenly more capacity is needed.
And I think that more power is needed to start a motor up than to run
it at a constant speed.  Which leads to the same question.

Can you direct me to a site or sites that really and clearly explain
this?  I hope so but I am pessimistic because so often the net does
not have such information--it is the kind of question (as are so many
questions I come up with) that you might find in a textbook or better
with a knowledgeable teacher.  If anyone feels similarly perhaps you
can commiserate with me over the loss of the Askme.com site.

I'm hoping that some readers of this question will succumb to the
impulse to add his/her knowledge.

This is actually a complicated topic.

At the root, there are generators that spin to produce electricity. 
The faster they spin, the more power they produce.  They can't change
their rotational speed instantaneously, though.  What happens is that
when a load drops off, the generators continue to spin at the same
rate.  This causes the system voltage to rise (extremely slightly),
which actually causes everyone else to draw slightly more power (for
resistive loads, the power is proportional to the square of the
voltage).  Eventually the generators slow down, and the voltage goes
back to normal.

This is a gross simplification, though - I skipped these details:
  - there are actually thousands of generators, connected in a complex
grid.  They all affect each other, and aren't necessarily tightly
coupled.
  - there are voltage regulators, etc. all through the system which
also affect the voltage, and delivered power
  - the line frequency (nominally 60Hz in the US) depends on the
generator speed.  The line frequency actually does move when they
speed up or slow down of course, but the effect is very small.  The
peak deviation is usually under 100 parts per million, and is designed
so it tends to average out over a 24 hour period (load is high during
the day, less at night).
  - this works pretty well in the steady state, when there are
millions of loads.  In this case it averages out very well, and the
instantanous load actually doesn't change very fast (variance is low).
 After an outage, however, things are much more complicated, and
manual intervention is often necessary to get things going.  This is
one reason why it took a while for the big US blackout a few years ago
to end- it was non-trivial getting generators synchronized again.

So even between day and night use, the average doesn't change all that
much?  Or does it change MORE than just a little but the system easily
accommodates.

Can you put your finger on a general amount of energy needed in
daylight hours versus night hours?  I mean at 7 AM West Coast industry
would really be getting into gear, but it would be only 4 AM on the
East Coast who would, presumably, still be shut down (at least those
that don't run 3 shifts).  And of course in 3 hours the East Coast
factories WOULD start up.  I was assuming it was quite a bit. Or is
all that just a minor energy-demand change when you considere the
entire US and I suppose, Canada.

What an incredible and fascinating system!!  It was exciting to read
your answer and I thank you.  I am always stunned by the science and
technology and the wonder of our world!

Um, that answer is totally false.

For starters, your assumption "I understand that such power is used
instanteously as soon as it is generated"  is incorrect. Power is in
fact stored in many different ways, ranging from batteries to massive
capacitors to most commonly flywheels. The previous poster's assertion
that power draw is essentially constant is very false. "If you can't
dazzle them with brilliance, baffle them with bull****" is a phrase
that comes to mind, no idea where he came up with that stuff.

Here's a starter:
http://en.wikipedia.org/wiki/Grid_energy_storage

PS in general don't trust wikipedia =)

Do you have any statistics on how much grid energy is being stored
compared?  From your own wikipedia link:

For the most part, variation in electric demand is met by varying the
amount of electrical energy supplied from primary sources, usually
hydroelectric power plants and natural gas-fired turbines.

My impression is that storing energy is still a very niche thing - the
vast bulk of energy is produced at the same time it's consumed.  Do
you have any cites that show it's even as high as 1% of the total? 
I'd be very surprised if it were anywhere close to that in the US.

I never said "power draw is essentially constant".  I said it doesn't
change instantaneously.  There's at least a 4X change between night
and day typically (in fact some utilities have a real-time graph of
power consumption, like this one: http://www.eirgrid.com).  However,
at the generators, the output power doesn't change instantanously - it
takes a finite amount of time to change the rotational speed of huge
generators.  In the meantime, the voltage rises or falls to keep the
power constant, and the frequency varies - it's the only way a 
rotating turbine can change it's power output quickly.

Here's a description of generator (turbine) behaviour with a step load change:
http://www.pondlucier.com/TIPS/2002/February%202002.pdf

Since the generators generally see aggregate loads summed from many
thousands of customers, the typical step change from second to second
is pretty small for each turbine.

The concept is correct including saving of energy noted has to be
generated. similarly when the stored power is regenerated.The concept
can be under stood simply as below for AC power. when more than
necessary power is generated the requency increases. this rise in
frequency consumes the rise in power. viceversa when the connected
load is greater the frequency drops and results in indirect load
shedding. In actuality the tendency to reduce the frequency is sensed
by governors and either load is reduced or more load is taken up,
mainly by hydro power plants. Thus a sudden overload beyond system
capacity results in black outs. but things become more complex as we
get transmission lines and many other issues like equipment protection
and load flow changes

www.ieso.com

or imo.com gives the daily load changes predicted, actual loads in ontario canada.