The waterblock design section of the forums at procooling.com has very in depth discussions on this.
http://forums.procooling.com/vbb/forumdisplay.php?forumid=37
You made the analogy of air to water, keep going with it. The most successful heatsinks are designed with surface area as the most important concept. Why does a heatsink work in the first place? If you took a 3-inch copper cube and sat it on top of a processor, you have greater surface area, right? So, this should help out with the cooling... but this wouldn't be very effecient because there is hardly any turbulance created on the smooth surface of a cube. Fins, pins and grooves! Make the air move around, through and over the copper and you have better cooling properties.
Take that same view and put it into a contained space. Now swap out air with water. If I had the same 3 inch copper cube, except it was hollow and had 2 holes in the top... you have a water block. Water goes in, water goes out. But, like the air analogy above, the cooling sucks. Add some fins, change the input/output locations, add TURBULANCE... the water then has more chances to hit the metal and cool it down. Bingo, a more effecient waterblock.
Doing that was the easy part back in the day... everyone could just slap a copper shroud around a normal heatsink and have a working waterblock. Lots of time spent and effort have been spent in research and development on creating just the perfect amount of turbulance, resistance and surface area... and a lot of great waterblocks have been produced. If you look at the old waterblock designs and compare it to now, you will see the results of all that research.
Better turbulance... more surface area... better flow design... it all adds up to better cooling.