After finding this article, I realized that would be an impossible task without an engineering degree.
Has anyone found a more "For Dummies" version of this?
If you want to really learn by building, I recommend "Fundamentals of GPS Receivers: A Hardware Approach" by Dan Doberstein, and building the analog GPS system described in the first few chapters.
If you simply want to play with GPS, I'm sure there are many breakout modules from Adafruit et al.
To be fair, that would probably be navigation by naturally observable indicators such as the rising or setting sun, shadows over the day, locally / seasonally reliable stars for your area, and knowledge of local fauna/wave patterns.
After getting interested in sailing again (after 20 years) last year through multihull vessels, I started reading about the evolution of multihulls in the Pacific and then began looking at traditional navigation techniques. Following that, I was wondering for fun whether it would be possible to plot one's position given a rough time, rough position guestimate, and perfectly measured horizonal inclination, lens property and visible to 'the naked eye' star positions (ie. visible to the naked CCD at evening time = high ISO, noise and short shutter speed to failitate clarity in rough seas). My conclusion was yes, it's possible, with sub-GPS accuracy but accuracy which would be good enough to get you reliably between certain locations in certain seasons even on a sailboat.
I was surprised to read that some ICBMs allegedly used astronomy and pre-loaded landform navigation to avoid issues with jamming, and that fat databases of precise naked-eye type visible star locations are widely available, indeed so is software for reducing them to identities from an image input.
(The big disadvantage is that it can't see through clouds, of course)