I'm already using two 12dbi antennas inside my car, "looking" through the side windows, yet the Mobile Mark 5dbi on the roof still seems more sensitive.
That''s not surprising. Any antenna outside the car will do (potentially much) better than an antenna inside the car. There may be (at least traces of) metallic content in the automotive glass, and assuming you're car isn't non-metallic (plastic, fiberglass), the rest of the body will cause shielding and all sorts of nasty multi path within.
my understanding is that the whip has to be cut to some length that is a multiple of the wavelength of the frequency one desires to receive
That's correct. There are many, many designs. 1/4 wave, 1/2 wave, 5/8th wave, and tuned elements can be "stacked" in-phase to compound gain, or perform on multiple bands (this is why you see wound coils in the element itself, or coupled between elements on commercially available antennas.
Rather than have the "porcupine" effect (4-16 whip antennas), "the CUBE" would revert to 16 plane antennas to keep gain high, but to use an array of directional (high gain) antennas, to achieve omni-directional reception. This is done all the time. This photo shows the master FM array on Prudential tower in Boston (note the "grappling hook" looking things, each one is an FM antenna). By stacking so many elements (mind you, the placement is precise, as to remain in-phase), massive amounts of gain is achieved. Also note, that there are elements pointing in each direction (N, S, E, W) around the sides of the tower.
http://gallery.bostonradio.org/2004-05/ ... 4-med.html
Since the horizontal pattern of the 2.4Ghz plane antenna I was looking at is 80 degrees, a similar configuration will work for "the CUBE".
I'm not sure if you saw my other post, but I did analysis on the first 90,000 or SO APs I found. 97-98% of them were on channels 1, 6 or 11. Since I've reconfigured my current setup to "lock" 3 cards, 1 each on 1, 6 and 11, with the fourth card hopping among channels 2, 3, 4, 5, 7, 8, 9, 10, 12, 13 and 14, I've noticed that I'm picking up many more networks (and geometrically more AP/position combinations in Kismet's .gpsxml file) based on my file sizes and number of unique networks per file set.