The Tweet-a-Watt has inspired me to come back to an old project I did some research on a few years ago - power monitoring with an AVR.
The main problem is how to safely measure power on the mains in a safe way. The Tweet-a-Watt idea is genius in that respect.
I was thinking of using a Hall Effect sensor. You get easy connection and isolation from mains voltages, and can use the ADC after a bit of scaling with an op-amp to measure current. The problem is that you also need to measure voltage to calculate power. There are some people doing that with sites on the web, but I had a different idea.
Maybe I'm just spouting rubbish but since the mains voltage is basically a sine wave with fixed frequency (50/60Hz), with a bit of maths you should be able to estimate power pretty accurately without actually bothering to take voltage measurements. At most all you need to know is the zero crossing point, and if you can't infer that from the current measurements then at least it's easy to measure and isolate with a simple pulse output from a transistor fed into an opto-isolator.
I'd love some feedback on this idea before I go and blow myself up