Vampires can be tricky...

Yeah those vampire loads can add up.  Identifying all of the vampire loads in ones house can be a daunting task, with the hardwired loads (like your ceiling fans) being especially difficult to detect.  Then of course detection is only half the battle, doing something about them then becomes the challenge.

If one is (really) serious about at least detecting these kinds of loads, one approach that could be applied is as follows.  First turn off everything in the house.  For devices that operate automatically, such as refrigerators, either unplugging them or at least adjusting the thermostat such that it is not likely to turn on while these tests are being performed would help.  Then turn off every circuit breaker in the house.  Observe the power meter and note that the spinning disc has completely stopped.  If not, there is still some power being consumed downstream of the power meter.  Obviously this approach will not work if your power meter does not have some visible indication of power flowing through it, such as with an all-electronic power meter.  Then turn each circuit breaker on, JUST ONE AT A TIME, and again observe the power meter for signs of energy consumption.  Timing the spinning disc can give a relative indication of how much energy is being consumed per unit of time.  There are actually formulas you can use to determine exactly the amount of energy being consumed based on how long the spinning disc in the power meter takes to rotate.  There are even whole-house power meters that attach to the electric company's power meter and can wirelessly communicate the power consumption to a receiver that will display the information as it comes in.  But simply speaking, the amount of power consumed (i.e. watts) is inversely proportional to the amount of time the spinning disc takes to rotate (one full rotation for example).  Unfortunately very small loads will cause the power meter's disc to spin very slowly, making it a lengthy process to time one rotation of the disc.  But this approach can give you an idea which circuits in your house have vampire loads attached to them, and also how large they are.  Granted this is a fair bit of effort to go through but the upside is that hardwired phantom loads will not escape detection this way.  Another downside is that after it is all over you will have to reset all the various settings on "always on" devices such as clocks and VCRs.

Once an individual circuit in your house has been identified has having phantom loads worth looking into further, knowledge of which parts of the house are supplied by the circuit is needed.  Usually this is written on the inside of the door to the breaker panel, but is only "suggestive" in the sense that you still may have to investigate further to determine exactly which outlets and light fixtures it supplies power to.  If you have one of those plug-in power meters (Kill-A-Watt for example) it can be used on anything that plugs in to get an idea of it's phantom load, although this kind of meter is fairly inaccurate when dealing with very small loads.  And such a meter won't help with hard-wired loads.  Unplugging each device plugged into a circuit under investigation, and again observing the spinning disc, while a time-consuming task, can help get to the bottom of where phantom power is going.  Taking notes may be useful if a lot of devices are involved.  When all devices are unplugged from a circuit under investigation, and the disc is still moving, you know you have a hard-wired phantom load.  Further isolation will rely on specific knowledge of your house wiring so I'll have to stop there.

Fixing the problem will of course depend on what kind of load has been identified.  Some are relatively necessary such as the clock on a VCR if you often program it to record unattended.  Others are easily turned off by just unplugging them or running them through some kind of switch like a power strip.  Obviously work to fix the larger phantom loads first.  "Fixing" some phantom loads will involve a tradeoff of convenience versus energy consumption, as in a rarely-used-to-record VCR, leaving it unplugged most of the time comes with the slight inconvenience of resetting the various options and the time each time it is plugged in to record something.  So something like that will be down to a personal decision of how much convenience is worth how much energy consumption.

Keep us posted!  I have yet to successfully convince my landlord to replace our porch light with an LED.

I have some low-voltage outdoor lighting fixtures which use incandescent bulbs similar to what you are describing (12 volts, wedge base, 168/194 style as used in automotive applications).  The orientation of the socket is such that the light needs to be emitted at a right angle to the socket.  I found these and am having good luck so far:

http://www.ledlight.com/detail.aspx?ID=255

They're just a few watts each and put out way more light than the incandescent bulbs they replaced.  Be warned that they apparently are NOT rated for outdoor use, I have had to replace a couple of them over time due to the elements eventually causing them to fail.  So not an ideal solution costwise if this kind of replacement is frequent.  I don't know how exposed the inside of your fixture is.  But there are many similar 12-volt LED lamps with the same kind of base on that site, some that emit light in all directions, and which also look like they could stand up to the elements.  Another source of this kind of LED lamp is here:

http://www.superbrightleds.com/mini-wedge.html

It might boil down to just trying a few that look like they might work, then once you find one you like, ordering enough to fill the whole fixture (which I guess is just two bulbs by re-reading your post).  You can probably reduce your energy consumption by 5-10x this way, and the appearance can even be better.  I know my outdoor lighting looks much better with LED light, and the energy consumption was cut from around 75 watts to more like 15 watts.