Well, in a nutshell, it's possible to combine multiple signals into a single waveform and then decompose them on the other end to get the original signals back, which is the general idea behind composite video. The composition/decomposition process isn't going to be perfect, though, and between that and degradation due to interference along the transmission path, there's information lost. That's not really an issue with a digital signal since it'll either work on the other end or it won't and there's additional data to recover the occasional bit lost or changed in transmission, but there's no way to have safeguards like that in an analog signal.
So, the alternative is to not combine the signals and just leave them separate, which is what happens with RGB. Rather than having one line carrying the information for the picture, you have four (red, green, blue, and a synchronization signal) that can be used with very little processing on the other end. Data loss due to interference still occurs, but since it's significantly less than what happens with the wave transforms, you end up with a much better picture.--Kreeblah
From composite (the yellow cable) to s-video is the biggest jump to me. Composite is blurrier, and worst of all, has this "dot crawl" effect that is especially noticeable on older consoles due to its flat graphics. Composite basically combines all the video information into one crap RCA cable.
S-video separates something called luma and chroma, and makes the dot crawl cease to exist. I officially became a videophile the day my friend begged me to take the s-video cables for his Dreamcast because he bought them by accident, and asked that I give him my composite cables since his tv lacked s-video.
RGB is named as such because it runs red, green, and blue separate. Since the image is so razor sharp with s-video, I don't see a huge difference with RGB. Most people say it is in the color reproduction.
You really benefit from RGB when it is outputting component video from it, and the device you are using can support a progressive output. For example, I notice a huge jump from my Wii when using s-video and component because the Wii can do 480p. It cannot do 480p over s-video.
However, s-video can do 240p just fine, so I really don't notice a huge jump from it to RGB. If using a CRT, I'd honestly stick with s-video because the whole scart thing is a pretty big pain in the ass. If I had a Sony PVR though, then I'd definitely go nuts with RGB.--Hamburglar