Range finder electronics exists as packages ready to integrate.  Just choose your distance, short is ultrasonic, long is laser.  Examples:

http://analogmodules.com/products/lrf-receivers.php

http://www.societyofrobots.com/sensors_sharpirrange.shtml

I know this is old but just ran across some Light to frequency converters. With a resolution of 16 bits and a wide range of sensitivities, Looks as if this might be the low cost answer. Please check taosinc.com for more info.The available devices include both analog and digital.Some with programmable gain. for a wide range of light sensitivity. Couple this with a reflectivity of block A or source on block A with sensor on block B, the range can can be varied. Also using an adjustable light source, the distance can be varied over a wide range. The cost of the sensors are only a couple bucks. I am going to build a motion sensor using one of the analog as well as one of the digital sensors. I have a need to measure the absolute speed of motion of an object independent of direction. Sorry for the long response time.

Bill

The range for capacitance measurements depends primarily on the geometry, of which the distance is a big part. I have found that the usable range is about 1 to 3 times the diameter of the electrode "plate" used. In this case, one plate is the center sensor, while the other is the ground ring, complicated by the "reflection" off of the Block B target. With sufficient width of the ground ring, one could expect decent results to about 2 to 3 times the sensor diameter. If you get much farther than that, the amount of gain required is so great that the noise become problematic (although you can recover from it to a point if you can average over a long enough time).

Also, note that Block B does not necessarily need a foil, if it's dielectric constant is sufficiently high. E.g., the raw material used in soylent green is full of water, so you can use this technique as a decent "person detector".

String pots aren't a bad idea, except when you have to worry about humidity changes and string stretch.  Good for being able to replace them though, and wouldn't be too hard to implement a centering capability with end cap limit switches.  The LVDT's are quite a bit more accurate, IMHO, used them for flight control position sensing on Helicopters.  Currently work with encoders and wheels that measure speed/distance for conveyor belts.  They work well if you can find the right combination to get the 7.5 degree measurements to equal a specific linear measurement.