In the arena of precision surface grinding is where I've found a good surface plate and gage blocks to be most useful, and mostly indispensable when doing that type of work. I recently spent several years making precision gages and CMM fixtures for a tier 1 automotive injection molder and learned to uses these tools to accuracy about as far as I believe one can go using mechanical methods. We regularly worked in tolerances down around +/- .00005" and these tools are the only method I've ever seen for consistently producing results at that level.
All of our surface plates were Starrett black granite and were professionally checked/calibrated every six months. We had a few pink ones floating around, but they were individually owned, usually in smaller sizes and riding on someone's toolbox for use during machining. The laboratory-grade plates we used were flat within .000035" across a 30" diagonal plate (or smaller) according to Starrett. The calibration service we used claimed the plates were resurfaced to at least factory specs and usually just a little better. Of course, all the tools I'm mentioning here were used in a very clean environment that was kept at 72°F year round. It was important to never directly touch any of these tools, as your body temperature would be transferred into the tools and cause them to grow ever so slightly. No resting of the forearms on the surface plates, no breathing directly onto the tools when doing close work, gage blocks handled with large tweezers, etc.
In practice, we would be given a gage component that most typically needed some part of it (either a physical feature or its location on a gage plate) altered. We built these gages internally according to a CAD model provided by the manufacturer, but most finished gages would require some fine tuning after being checked against the model used for CMM verification. I always wondered why we couldn't use the CMM model to design the gage in the first plate, but I'm not a CAD guy and never could get a straight and coherent answer from anyone as to why it was done this way. Anyway, let's say a step machined into a block needed to be moved down .0004" (Z axis) and over .00125" (X axis). First, I would clean my surface plate with surface plate cleaner (no mineral spirits, acetone, Windex, etc.) and the part with acetone. I would go through the motions of "wringing" the part to surface plate. Although the surface finish wouldn't allow the actual wringing to occur, this usually provided the best contact between part and plate. Calibrated carbide gage blocks were built up to match a control surface, such as the top of the part, and the current height of the feature I was measuring and these heights were verified using either a .00005" or .00002" dial indicator mounted on my heavy Starrett surface gage (p/n 57D, I believe). Once these heights were verified to my satisfaction, I could re-stack the gage blocks (for the surface I was moving) to the desired finished dimension and proceed with grinding.
After grinding, the part was cleaned and again wrung to the surface plate. After verifying my control surface and corresponding gage block stack co-witnessed, I could check the newly-ground surfaces against their corresponding gage block stacks and say definitively that, in this example, the surfaces were moved precisely .0004" down and .00125" over. This info was documented along with the control numbers of the tools used to verify, and sent to the quality assurance lab for final verification and return to service.
In addition to all these precision calibrated carbide gage blocks (in the neighborhood of $8-9000 USD a set!), many shops use master height gages, such as the Mitutoyo 515. Good, used examples can be bought all day for less than $1000. Combined with a really good surface plate, you can get pretty deep into serious dimensional accuracy for a minimal outlay of cash. Naturally, cheaper plates and gage blocks won't offer extreme accuracy unless they're properly reworked and calibrated, but they will provide you a way to do more precise work than what can be turned out using more common methods. I would have a difficult time getting by without mine, but then I've been using them for quite a few years.
After all the entire hobby by definition is not a need but a want
