Tom, would that help if I'll explain that I've been serving in the Soviet army? Where most of the problems are resolved with a sledge hammer or with just a strong taboo word magic...

Let me explain the obvious to me "prismatic" M73 compass (P) flaws vs (N) $5 noname compass advantages as well:

1. P uses a prism on a hinge, N does not have any prism nor a hinge there (so the N is mechanically more reliable and less expensive to manufacture).

2. P's prism obscures the dial when engaged, N's dial is always open.

3. P has a fixed lens after the prism, N has an adjustable lens (great if your vision is not 20/20)

4. P has an eye sight slit on the prism's body. If you bend or damage the hinge you will have a wrong sighting if any at all if it lost (the prism's internal hairline will be shifted as well). N has that sight fixed on the monolithic compass body.

5. The P's prism mechanism has no hard stop in the engaged position (over the dial's glass), it is possible to have it engaged only half (or even if 7/8) way and have the measurements done without even noticing that. But in case of just slight misalignment of the hinge you will have a systematic error introduced. Yes, they are tested at the factory, but only in the 100% engaged position.

6. The dial of the P has their most precise ring numbers mirrored (prism is mirroring that image back to normal) - that makes it less readable for a direct view (asking for silly human error). N has totally normal numbers.

7. When you disengage the prism on P it's protruding far from the body making it vulnerable to various forces. N lens is screwing almost flat into the body.

8. P eyepiece lens is mounted almost on the surface (maybe 2mm deep maximum, and beveled out, why?), N lens is deep (about 5 mm) in the holding black tube (what makes it almost entirely glare free and protected from any abrasion chance).

9. P dial's degrees lines are obviously radial, N's are much closer to parallel being on the vertical slope (what makes it more reliable when taking measurements for fractions of a degree).

10. P's dial as visible from the eyepiece has no information about the levelness of the compass, N's side view reveals even slightest slope in any direction immediately.

11. When used in direct view mode, P has no bubble level and it is very hard to see if the disk is level or not from the top (easier with the needle - true). N has a dedicated bubble level on the top.

12. P and N have exactly the same front sight, which is a transparent window versus typical non- or semi-transparent mirror with a slit in non prismatic alpine compasses (that's the answer to your question about the mirror).

13. P has round body, N - square (what is better for map work?)

14. The tripod mount is great to have than not to have it at all. Perhaps, you could understand its utility better if you can imagine replacing the "tripod" word with the "sturdy and level table" words? Have you ever tried to map a terrain on piece of paper with an ordinary compass? I did, and I did the same with a theodolite on tripod as well.

15. There could be more.

Regarding the typical baseplate compass measurements precision. I meant the compound accuracy (mechanics, techniques, conditions). I said it is 10 degrees. You are insisting on +/- 3 degrees, but that's 6 degrees overall. Add dial/needle/sight view parallax errors, add lack of levelness, add impatience for needle setting - and you can round it up to 10 without a single doubt (I'm leaving magnetic anomalies and lack of mag.dec. precise knowledge outside of the scope). And that's exactly the ancient compass technology flaws, you have mentioned yourself, as the primary source of these errors.

Also I don't like the "skilled hands" notion. I understand the pun and paraphrase behind, but perhaps you don't? The skills here are is the deep knowledge of your tools and of the proper ways to overcome, amend, correct your data for any flaws they have and which might influence specific measurements. No mill.spec or the price guarantee you don't have any (see the list above). And I have stated at the very beginning that in the skilled hands all that matters in a classic compass is the concentricity of the dial with the needle pivot. The rest are just skills. However, the compass I've presented above is the improvement over the classic (ancient) model (the M73 is too, but the N is way better, because it's not struggling to be a Swiss Watch, but rather a very well thought out pocket compass). You can find the exact same scheme and features, which you can get for $5, in multiple models of $800 survey compasses. But there is no a single reason you have to pay 160 fold more to have all of that tech advancements, eliminating or minimizing many classic model flaws, for an ordinary field navigation.

The longevity of the design? Millions of happy users? The digging stick is much more reputable than a shovel then.