If I'm not mistaken, 20,000 units, at $500 a pop, is roughly $10million--in a week!--with hardly any overhead (ok maybe a million here, a million there for online advertising).
That aside, Google's approach from the beginning was never shock and awe. It was and still is more organic. Maybe Google is taking a playbook from the 1980s Apple v. PC playbook, where Apple remained mired in a closed platform and lost. Apple came out with a bang in 1980, but quickly lost ground to PCs based on MS-DOS. The first PCs got the job done, and some sold better than others (100,000 units in the first week versus 20,000?). Yet they made up the gap with overall volume and market saturation (think Hero, MT3G, G1, Moment, Eris, Droid... N1). Here we are 30 years later and what variety can you get from Apple? Yep, a closed source iPhone.
In my opinion, Google's goal with the N1 wasn't an iPhone killer, it was a revolution. And revolutions don't always change things overnight, or even over a week; sometimes, they takes months or years. 20,000 units is--in the grand scheme of things--just a number, but it is 20,000 more revolutionaries in Google's pocket (or vice versa?).