I was hoping that the link would say that they can't read PGP encrypted messages or something along those lines. Alas, we can't be too sure about that either.
Thats what I thought. We had a few teams look at ways to carry human voice with less BW in a sensory communications course. The winning team did it with 500Hz. I remember it clearly. There was a stunned murmur during their project presentation when they announced their results.
I don't remember the exact numbers, but the gist of it was, they applied 2 bandpass filters, one on a lower frequency range and the other on a higher frequency range. Both passbands were 250Hz wide. They combined the two signals together as a 500Hz signal. The pass bands were seperated by 500Hz or 1kHz IIRC. So pretty much they kept the lower and higher frequencies and threw away the rest. The result was pretty good. Human languages have a lot of redundancy in information and the quality was good enough for general conversation. It was comparable to what you have on a cell phone.
If you are looking for something that you can run from a USB cable, this could be the perfect tool. Someone called this the Bus Pirate of RF and I think its a great way to describe this tool.
The SDRs from Per Vices are geared towards low latency and high bandwidth. You can't squeeze 4x PCIe bandwidth into a USB cable. Neither can you get sub microsecond latencies. You can't get this over an ethernet connection either. This is the application where the Per Vices SDRs come out as clear winners.
The question of receiving signals is more a philosophical one of ethics. The question of broadcasting signals is one governed by law. In North America, you can't broadcast on a part of the spectrum that you don't have a licence for. Wifi devices, 900MHz phones and CB radios and such operate in a frequency band which has special provisions for such use.
I looked at their product specification and there isn't any mention of phase noise of the local oscillator. It turns out that a critical requirement for tuning into GSM signals is the phase noise or jitter of the local oscillator. Chances are, they are using a pretty good cystal, but probably not good enough for a GSM base station.
There is a good reason why GSM base stations all run off rubidium standards. In fact, you can grab 10MHz rubidium standards for a very (relatively) cheap price off eBay. They are all from decommissioned cell towers.
There is a possibility that you can still tune into a GSM signal with a substandard local oscillator. However, this behaviour would be intermittent.
Depends on what you want to do. A lot of people think that they need more bandwidth or power to achieve something. This may not necessarily be the case.
QRP guys have crossed continents on less than a watt. I remember from my undergrad days that we've played with band passing the human voice. We managed to get surprisingly good results with 500Hz of bandwidth.
There's a lot of interesting things you can do with more power and more bandwidth. To echo a lot of the other comments made already, if you are doing this without much thought, this is generally frowned upon.