+1 Tim. I shudder to think what happens if we stop buying the expensive stuff.  
@ three_easy_payments, funny you should mention that.  My friend just bought a PrimaLuna from Upscale.  He's very pleased with it!
I agree with the article but there are some miscalculations in it.
For example: if you look at the # of bits per second that are read from a music file, they are around 800Kbps. 
Timing is everything when dealing with transfers of any data. If a packet of bits doesn't end up at its destination, the packet will be resent X amount of times and then sends an error. 
I used to design/tune/troubleshoot enterprise networks. In the old days with slower wired network speeds, you had many more issues of lost packets than you do today due to high lantencies. But, even today, a wireless connection can have many lost data packets over a certain period of time due to interference. Another example for tuning your network connections, try using jumbo frames, much less overhead than having your network devices divide up the data into 1500B packets. Test test test after making this change to verify if you see an improvement. If you don't, turn off jumbo packets.

CD players generate jitter and enough jitter can cause bit errors.

As for timing the dac uses, read some of Ted Smith's blogs/writings on how the DS dac handles timing.
The PS Audio Copper electronic mag is really great, well worth your time. Great subjects and writers.
It’s even more complicated than the article describes. The pits and lands of varying lengths on the CD do not (rpt not) represent the musical waveform. They represent predetermined discrete sets of digital data specified in the Redbook standard that must be further processed and assembled before the digital data can represent the musical waveform. There are about 10 predetermined discrete sets of 1s and 0s.