My understanding of E-TTL is also pretty limited, but from what I understand here is the basic idea behind it:
1. Take a shot while an E-TTL compatible flash (ex. your 430EX II) and the camera are connected in some E-TTL compatible way (ex. hotshoe, or hotshoe extension cord, E-TTL radio transmitter).
2. The speedlight sends out a flash.
3. The camera records a temporary image and analyzes the amount of light in that image. Your camera settings can allow you to set it to correct exposure for the whole image or the area that your auto focus is focused on. The camera then determines whether it needs more or less light and makes adjustments to correct the light on the next flash. Using the lens (and sometimes built in speedlight technology I believe) the camera also notes the distance the subject is from the camera. This helps it come up with a better calculation for how much it needs to adjust the light.
4. A second flash is sent out by the speedlight, this time with either increased or decreased brightness to achieve better lighting.
Everything from 1-4 happens in a fraction of a second so fast that you won't even see it as 2 flashes. To the human eye the first flash, exchange of information, then second flash, happens so fast that it isn't even noticed.
From my own experiences, you can't get accurate E-TTL adjustments when you bounce a light. If you are bouncing the light off of a ceiling or something similar i doesn't seem to be able to calculate an accurate adjustment. I don't deal with E-TTL very often so this is just from a very small observation that I have made from the couple times I have tried it out. I am also interested if I too was just using it incorrectly, or if bouncing light or using light modifiers just mess up E-TTL.