Not saying you should buy and you are right there is a lot of risk and even a small delay can mean a lifetime loss BUT those calculators somewhat idiotically assume exponential growth forever. If you take that calculator and extend it out two years it would be predicting that mining will use more energy than all other purpose by the human race combined. The electrical costs would be something on the order of $340M USD per BTC. So not exactly realistic.
The calculator is only as good as the assumptions. Garbage in, garbage out. So it is up to you to adjust the parameters to what you think is realistic. At some point it will become increasingly difficult to keep doubling the network every month so changing the calculator to show a certain increase in difficulty (X hundreds of millions) per month is more realistic.
I'm not certain of the implications of that scenario. If network speed can't keep up with the difficulty increases, then the difficulty drops right? It seems like this scenario would only happen if people stop mining. Maybe I should look into building my own ASIC miner instead...
