OK, let me put my geek hat on for a bit. The coordinates in the GS book may or may not be GPS derived. Remember, GPS has to do with the satellites and the receiver. I would guess many of the "bad" coordinates are from a surveyor who generated them maybe 30 years ago. With modern GPS, if you go park your GPS receiver at the office/gate/whatever, the latitude and longitude it gives you is going to pretty accurate (easily 10's of feet). Datum that SCVJeff talked about were used by surveyors and some early aerial photographic techniques and could account for some of the error. There is also the old, "I copied it wrong" error and the "I am a human and made a mistake in my survey" error.
To hopefully clear up a few things (maybe cause more confusion, I don't know): The earth is closer to a geoid than an ellipsoid but GPS uses an ellipsoid because it is easy to describe mathematically and it has a unique correspondence to the Earth-Centered-Earth-Fixed (ECEF) coordinate system the GPS uses for navigation.
GPS coordinates are based on the WGS-84 ellipsoid. None of my GPS units allows me to select a datum and nothings I have seen leads me to believe these GPS translate the latitude and longitude into any other system. Joe surveyor (or campground owner) may use his GPS to determine a location then have the GPS translate them into a different datum (this is additional software), but I do not see anything that indicates RV/car/Truck GPS systems do this translation.
GPS assumes the earth is an ellipsoid then uses a geoid model to determine the Mean Sea Level for a given location and an Earth Gravity Model (EGM) to generate the terrain elevation. The terrain elevation (where the dirt ends and the air begins) was generally derived by Shuttle radar measurements but may come from other sources too. All the none "average elevation" stuff is easily accounted for.
Virtually all GPS receivers use the same ellipsoid to determine latitude and longitude. I do not know of any GPS that does not use WGS-84. In addition, with modern GPS, two receivers should give virtually the same location even if they are not tracking the same satellites.
GPS accuracy has some dependence on your location on the earth, more correctly your location relative to the GPS satellites but for most of our travels, that is negligible.
A note about precision and accuracy, do not confuse the two. In this application, precision is more or less how many numbers are to the right of the decimal point. For instance, 1.001 is less precise than 1.00100000000. The one with all the extra zeros is more precise. Accuracy is how close you are to the correct answer. If the true answer is 1, then they are both inaccurate by 0.001, all the extra zeros just means you are more precise in your error. That is the difference in a nut shell. As far as the relationship between precision and accuracy, there is obviously a connection that if you do not have enough precision, your accuracy will suffer. In other words, if I put things in whole number miles, I will have a hard time getting anything accurate to within feet. In the world of latitude and longitude, the number or fraction of feet represented by the digits in a number changes over the earth. In my home town near St Louis, MO, a foot north-south and east-west is represented by approximately 0.000003 degrees. In other words, the position 38.842266, -90.315859 is about 1 foot north and 1 foot east of the position 38.842263, -90.315856. So if you add more digits to the right of the decimal point (e.g 38.842266454654541), you are adding inches/fractions of inches to the position which, in our case, has no value in accuracy.
Since I am retired I no longer have a proof reader so I apologies for my math, format, grammar, typo and spel'n errors.