At the current rate, probably would take you a year at the least, not including power. difficulty increases but also decreases supply. If we hit $30/btc again, it'll be worth the investment, then again if btc tanks tomorrow, it wouldn't have been. It's a gamble.
Hit it right on the nose - mining is kind of like playing the lottery; it can pay off, but the vast majority of the time you pay out instead. Even if there is a massive drop in difficulty, the number of coins generated over times drops, so your investment loses value the more people know about mining and attempt it. Even in the case of a phenomenal drop (you were one of 5 miners), the difficulty would adjust automatically anyway over time. Long term, mining is profitable under the theory that you run a dedicated miner for many years, and only assuming that the number of miners doesn't increase TOO rapidly in that time.
A best bet option for mining is to contribute unused CPU cycles from existing unused hardware to a pool in terms of payout/reward, if your operating (electric/bandwidth) costs are low enough to justify the long term investment in BitCoins and it appreciated to a profitable level before you sold. If you are determined to buy a dedicated mining rig (if you have very low power costs, for instance, this might make sense), the same principle applies - make sure you are profitable per sale.
Many miners go about it the wrong way, mining at a loss and then selling BitCoins almost immediately to reap some reward. In order to be a profitable miner, you have to measure your costs accurately, set your profit margin, and sell when the price meets your goal. Anything less planned is usually just paying too much for BitCoins vs. market value. and getting a really high framerate in Crysis.
