If you believe that the value of 1 BTC is going to be significantly higher in a X years than it is now, then yes. 2 years ago, 1 BTC was worth about $1. So even if you were mining 1 BTC per day, it was "worth" about $30/month. That probably would not cover the power cost so looking at the then current value, it was not worth it. The argument then was that "mining was not profitable" and that was true
at the time but it excluded future potential appreciation.
Even at $5/BTC roughly a year ago, it might have been only break-even on power. At $120/BTC today, that changes completely what a machine was mining
then.
So, if you believe that the value of 1 BTC will continuing growing (whether or not at the same rate as in the past 4 years), mining 0.1 BTC/day (or even 0.05 BTC/day) could be worth it in a few years (even before the next block reward halving in ~2017). Looking at the cost of power today and the value of 1 BTC today alone would lead one to say it is not worth it, but it ignores future appreciation potential.
In short: if you intend to mine and sell today, then you probably (definitely?) will be losing money, but still may be fun. If you intend to mine today, and then spend in 12 months or 48 months or longer, then it may be worth it.
You can always use the rig for LTC if you want later.
I've always wanted to build a multiple video card machine but never had the reason until now with Bitcoin mining. I mean, I never used cross SLI support with multiple video cards. Something about stuffing 4 video cards into one machine and overclocking them sounds fun.
But I feel like I'm too late to the party. Searching for the cost efficient graphics cards is proving difficult as they are out of stock or higher in price due to the demand.
Any suggestions? Should I wave to the boat or jump in the water and try and swim up to it?