It's too bad the author gave up on a Mac install. If all you need is the low level fundamental C++ libraries of GNU Radio, it can be done without installing dozens of support libraries. It's when you try to get all of the Python and graphic stuff happening that it gets really complicated.
A Mac Mini with one of these $20 sticks is a decent little system for working on this stuff. It has enough horsepower, is totally quiet, doesn't take up much room, and only uses about 20 watts during ordinary receiving tasks.
Feel free to ask any questions about the rtl-sdr project. I've been watching it since it started and have since gotten a handful of commits into the project.
Some key things that were not mentioned. There is the GQRX frontend and the SDR# frontend. Both are fairly straightforward to set up and use.
Also not mentioned were the hardware differences (r820t vs e4000 vs fc0013). To get decent signals you need to add a bit of filtering to the USB power supply. And of course antennas, but that topic needs an entire book.
I've been working on a very simple sdr stack (currently merged in rtlsdr as rtl_fm) that is meant to provide a sox-like experience. It is also stupidly fast, enough so that even a little RaspberryPi can easily do pager decoding or police scanning.
Ferrite clips/donuts are recommended and easy to install. Put capacitors in parallel with the power, exact values vary. If you are feeling really ambitions, add ferrites in series with the power lines or give the dongle a solid metal enclosure.
The simplest is using a usb extension cable, the computer itself generates large amounts of broadband noise.
Over what frequency range can the dongle be tuned? What is its useful bandwidth, and is it limited by the radio/sampler or the USB interface? Also, what sort of sensitivity/noise figure do you get? Any comments on inter-modulation performance?
Lots of questions, but I'm curious as to how these receivers compare to a more traditional rig, and where the limits of performance lie.
A (relatively) simple hardware mod lets you bypass the tuner and pick up 0Hz to 28MHz.
Noise and sensitivity vary. I know at least one person ran a bunch of tests with a spectrum analyzer. (Can't find that link though.)
Performance generally seems to be limited by the poor shielding/filtering and the low quality timing crystal. If you fix the worst flaws, they are rather capable. For example, they've been used for GPS and radio astronomy.
Biggest differences with a more traditional setup is the lack of a transmitter and the questionable build quality.
Some may wince at the idea of assembling a kit (esp one with SMT parts), but if you're not scared by those factors, check them out. No ham license needed, though if you use the stock crystal for the LO, you better brush up on your Morse Code! (I suggest using the 28.59MHz xtal)
Somewhar off-topic, but interesting: you can buy some really complicated-looking radio gear. For those of you out there who love stuff with a lot of switches, check out this ICOM radio.
I have no idea what it does -- I'm assuming it's very similar to my son's ham radio -- but it sure looks cool. Hobbies this complex have always been intriguing to me.
The coolest thing about the hobby is that it scales to all sorts of different interests and spending levels. Compare this to <http://smallwonderlabs.com/Rockmite.htm>. Sure, it has no buttons at all, but you build it yourself and you can still have a wireless conversation with someone half a world away.
A Mac Mini with one of these $20 sticks is a decent little system for working on this stuff. It has enough horsepower, is totally quiet, doesn't take up much room, and only uses about 20 watts during ordinary receiving tasks.
I took notes on how I made this work on my Mac here: http://rachelbythebay.com/w/2012/09/19/brute/