I'm hardly an early adopter, joining up 9 whole days ago. What made me sarcastic and condescending (not to mention bitter) was answering this exact same question about 30 times.
If we assume the difficulty will grow by 30% every 2 weeks (both of these estimates are very conservative, by the way), 700 Mhash/sec worth of hashing power will probably find about 1.95 blocks between now and the end of time.
You'll have to plug in what you think a bitcoin will be worth to find out if this is a good idea or not.
You'll need to scale that to the GPU you are planning to buy. If, for example, you get a 5850, you can expect about 300 Mhash/sec out of it, so it would earn about 1.95 * (300/700) = 0.835 * 50 = 41 BTC. Even that estimate is probably very high.
For 700Mhash w/ 30% difficulty increases I get about 70 BTC. Of which 60 are gained in the first 70 days. Even at difficulty increases of 0% the cost of a machine will take 40-60 days to recoup, and the cost of only adding graphics cards about 10 days, the thing is in either case the profit is so small and risk so high its not worth it IMO.