The biggest problem is the accumulated errors in our system will compound rather than average out.  

I take it is being inferred more elements should be included in the soil analysis.  Problem is the usual deviation from true for other than pH & OM is large for P, K, & Mg. The usual deviation for S & all the micro nutrients is excessive. It is difficult to put a lot faith into data that is most likely not representative of the soil.

There is nothing magical about the 2.5 acre grid other than it is roughly one hectare. A number of years ago the Journal of Productive Agriculture had an article. Their conclusion was grids larger than 70 ft by 70 ft has little advantage over whole field testing. In effect a 2.5 acre block will produce the same error as a whole field sampling. Add  to this work in Oklahoma looking at a one foot grid in a 7 ft by 70 ft block showed dramatic variation in P & pH. This exercise was conducted with an established bermudagrass stand. The was nothing magic about bermudagrass except it was easy to measure production in 1 ft² blocks by collecting clippings from each ft². As you would expect the production had a dramatic variation, but the productivity was not mirrored by the fertility.

As a test try double sampling a field. every spot a soil probe is pulled for test A, move over 10 feet and pull another probe for test B. Include them buried in a group of samples. More than likely the results will look like 2 different fields, grids, or plots.

Grid sampling is an interesting and attractive concept. GPS for sampling is also attractive. To insure uniformity the grid locations will need to accurate to 100 mm. I am not confident surveyors GPS instruments are that  accurate, let alone what we have available to us.

A number of years ago the aircraft industry adapted an INS navigation system. In a short time the aviation industry found radio navigation systems had sometimes several miles of surveyor error.  Even Airfield and runways were not where the coordinates said they were.  

With sampling errors, location errors, lab errors it is not impossible to realize a 30% or 40% error in the reported values. Throw in a lack of data for a local soil and any recommendation can become a SWAG.  Better than nothing but not all its is touted to be.

Each farmer, should know and understand their local soils, (plural). Most soils will show a 1 ppm P soil test value change with maybe 15 lbs/A P₂O₅, Soils with a high pH may see a 1 ppm P change with 25 lbs/A P₂O₅. With my more calcareous soils there is a 1 ppm P change with each 71 lbs/ P₂O₅. Something a lab less than 100 miles away may not be aware of. For sure a small detail that probably would be foreign to a lab 500 miles or more from my gate.

Still every lab in the country will insist their data is fully accurate not just the most accurate. Recommendations using a Big Chief pad or a Dell are of little value, unless there is a local close and personal knowledge of each soil.