... GPU mining is still fairly profitable with the half reward for ~most people (those who don't have electricity over $.15/kwh or so). My expectation is that the difficulty is currently lower than its proper equilibrium because people have opted to not purchase more GPUs while anticipating the introduction of ASIC devices.
Where do you get the $0.15/kWh figure from? I am curious, since this seems to make it sound like there is a stable formula for converting kWh's to BTC's. That will of course depend on the exchange rate, but I'd like to see the figures.
Why do you put a hardware cost of 0? If you put the true cost, let's assume $1000, it'll take you 273 years to make your money back. I assume the value of the currency will increase, but still... am I missing something as it seems a pretty flawed way to make money.
I put no hardware cost because I'm looking at the instantaneous profit function (or close using a 24 hour window) for hardware owned prior to the reward reduction. The comment was referencing whether or not it is still profitable to mine given a certain electric rate. Also understand that some people will continue to mine even at a slight loss during the winter for free heat, while others will sell their hardware and put the money to better use.
So, if you want to amortize the loss of hardware value over 24 hours be my guest. If you want to purchase hardware to (profitably) bitcoin mine, then (as you clearly recognize) you need to get the new specialty hardware not an off the shelf video card.
Bear in mind that most people's GPU mining rigs use cards that are a lot less efficient than the 7970, especially any remaining large-scale GPU miners.
... GPU mining is still fairly profitable with the half reward for ~most people (those who don't have electricity over $.15/kwh or so). My expectation is that the difficulty is currently lower than its proper equilibrium because people have opted to not purchase more GPUs while anticipating the introduction of ASIC devices.