Yea it's tough, probably not worth it. Basically, right now I get about 37.1 MH/s (with 3 GPU running) and make roughly $0.055 in revenue per hour, and if I give every benefit of the doubt, maybe someone can get 6 1070 GTX GPUs working at ~33 MH/s each for 200 MH/s on a 1250 power supply. So let's say someone can make $0.30 an hour before electricity as a max. Let's conservatively assume they pay $0.10 an hour in electricity, which equates to $0.20 * 24 = $4.80 a day, or break even in 1.5 years. Rough gig, but have to remember that the hardware has value and can be sold or run for a long time and if other coins become more profitable you can switch to those.
Just seems like a very long time to get to break even, and it's better to take your fiat and buy crypto directly than to mine unless you have either solar power or very cheap electricity -- Iceland or China.
I'd guess it's still popular because there are people like me who try it for fun, or people who overestimate how much they can make.
I want to build one for fun too but main reason is for profit. The more profit the more fun. That is just simple math here. Yet actually doing it is the hard part. Each passing day I do not go out and take a leap I will only feel I am missing out and the higher the difficulty.
37.1MH/s on 3x1050s actually sounds decent. I am likely going to pay more for new cards because I just don't trust the run time on the used cards. I have read many bad stories like the miners that were built and resold burn out in a few months because they were running of fumes. Resale values of those are likely dropping or will drop due to usage. I have been on ebay for two weeks and the rig prices are still ridiculous. Going to wait even though difficulty is on the rise.