In a nutshell, films are originally recorded at 24fps. However, television transmits at 30fps. In order to make film-based material run smoothly at the higher rate, 2 out of every 4 frames are used to create a new composite frame (this is the frame where you would see two ghost images, rather than one clean image). Where once there were 4 frames, now there are 5. (This is where 2-3 comes into play - 2 frames --> new composite made --> 3 frames altogether [the original 2, plus one extra] The next two frames are left alone, and the pattern repeats with the two frames after those.)
Frames
A B C D E F G H-- 24 fps
\...4..../\...4..../
A AB B C D E EF F G H -- 30 fps
\......5....../\......5...../
What IVTC does is remove that extra frame (which is essentially a copy of two other frames anyways.) Laserdisc requires that the extra frame be in there, because it's a simple analog playback system. DVD, however, is capable of many new tricks, due to its digital nature. Thus, instead of requiring the extra frame to be in the footage, we can use programs like Pulldown.exe to place flags where this extra frame should be. When the DVD player reads the flag, it creates the extra frame on the fly. Thus, our 24fps MPEG is decoded and changed into a 30fps video stream. Since the DVD player is able to extract and create this extra information from the existing frames, it saves us one frame for every 4 frames. When you realize that ANH is 170,000 frames, it turns out that encoding to 30fps creates @43,000 extra frames! More frames means a bigger file size. If we wanted a file size of 4 Gbs, then the more frames we have, the higher compression we must use. Higher compression = lower overall bit rate. Since we can get the same picture with 24fps or 30 fps, it makes more sense to save at 24fps, using a higher bit rate per frame, and make the DVD player do all the heavy lifting.
I hope that makes sense...