For that reason, I've been reticent about trying it. I've had an ongoing investigation -- a window-shopping habit -- to watch for compromises or hybrid systems that employ TEC. Apparently, the TT SubZero4G was something of a failure for the price, because resellers that once offered the gadget at $90 to $120 have discontinued it.
I was thinking along the lines of using TEC with a more elaborate heatpipe cooler. What that entails is a regulation of the fan and the TEC to maintain temperatures at some level -- preferably a level that stays above the dew-point with the condensation problem. Not exactly an easy project even for the "enthusiast."
Now I find that there is a company called Wintsch Labs producing a "Dominator-Pro" CPU water-block with a TEC cooler. Swiftech is selling a tiny device that uses TEC to refrigerate water somewhere in your water lines; one would imagine that it would allow eliminating a radiator from your configuration. Have to look into that more.
But the point is, with these sorts of solutions, as with AseTek Vapo-Chill, you begin to complicate your life with "heater-coil" and neoprene-gasket devices designed to mitigate condensation.
Suppose you use air-cooling to over-clock a 3-Ghz system to 3.6 -- or even 3.75. How important is it to you to get such a system to 3.9 or 4.0 for the additional cost and trouble? At least with a straight-forward water-cooling system that only goes as far as "water-at-room-temperature", you don't add other variables in the equation, you still need a fan for the radiator but not four or five fans pushing air through your case. And from what I've heard, there really isn't THAT much maintenance to worry about.
Even so, I wouldn't feel that easy about leaving a couple-grand's worth of equipment home alone with a water-cooling rig running. But that may be just a psychological thing. With air-cooling, you could lose a fan in your power-supply and really burn things up, or your CPU fan could conceivably go on the fritz and -- MAYBE -- your CPU would shut down (depending on your original choices and configuration).
What's the probability of failure, or equivalently, the MTBF for a water-pump as opposed to a CPU fan?