This one has me stumped:
from here
How does that work? Is there some kind of mercury switch (probably a fancy high-tech version of it) in the camera?
Yup.
Eh… on second thought, you were probably looking for more information than that. Sorry 
Here’s one example but the explanation is a little technical and doesn’t get to the physics of how it works; it just refers to accelerometers. That was based on a quick & dirty Google, there’s probably more out there.
Not a mercury switch like in your thermostat.
I notice my camera (Canon SD500) seems to make use of that information in another way, as well. If the camera is on but I’m just holding it down at my side, it will go partially to sleep (LCD screen turns off) after a minute, but when I lift the camera to take a picture, it immediately turns back on without my touching any buttons. It’s actually pretty cool and effective.
At its simplest, an acceleromter is an inertial mass on an elastic mounting, with a transducer to convert the amount of displacement of the mass into a signal. It exploits the Newtonian tendency of a mass to resist motion by remaining in place. Usinge MEMS (micro-electromechanical systems) technology, a modern solid state accelerometer can be enclosed in a package the size of a surface-mount transistor.
So, basically it’s just a kind of rigid pendulum?
Camera level:
------|------
|
|
|
|
#
Camera tilted 45° to the left:
/
/
/
|
/|
/ |
/ |
/ #
Camera edgewise:
|
|
|
-\
| |
| |
| |
#
(No! This is not a cow’s backside!)
And vice versa for clockwise rotation. The readout they get is the angle between the “ceiling” and the pendulum?