Andy T,
Re the Michelin Guide a few points.
The scale is akin to logarithmic; only a handful of anythings should achieve three stars, even as there is no bound on the number of anythings. This is unlike top 10s, top 100s, etc, where a completely arbitrary number is chosen, and then the appropriate number of anythings is pulled forth, regardless of whether there are X anythings worthy of note.
If you are having trouble differentiating *** from ** (or ** from *, etc), I would suggest you push courses downward until you feel significant gaps between your respective groupings. It's like the story of the scout passing through towns where no one can agree on the best player. (If there's no majority then no kid in town is good enough to make it at the next level.)
Which brings me to my next point about cardinal numbers and the Michelin Guide: there is no law that says there have to be *any* *** anythings -- or X number of Y-starred anythings.
The MG's red guide star system is intended to differentiate not all restaurants but just the very top tier. In truth there are more than three stars: making the red guide is notable in itself, and then there are further distinctions along the lines of good for families, etc.
If golf courses were reviewed as in the red guide, the starred courses would be an important but not the only part. There would be vastly superior information content relative to these lists of 100 courses. A real red guide for courses would identify the truly special in more respects, including "the best" in terms of absolute excellence (stars) -- but not just absolute excellence (whatever that is).
Which brings me to my last point: for many people a cardinal system makes for harder decisions and for better discussions. Are these two vastly different courses really on the same level? How does one assess excellence when making what amounts to apples and oranges comparisons? How are we to take individuals' views and convert those into a single value?
So we're forced to think more deeply about the nature of excellence, a good thing. Other than at a few arbitrary cutoff points (no. 10 vs no. 11, no. 100 vs no. 101) I don't think that really happens in an ordinal system. We just come up with an ordered list that lets us evade the hard choices of how much better no. 1 is than no. 2 -- or than no. 100 -- or of what we mean by excellence.
It is illuminating that these magazine rankings apparently have their rankers use some form of cardinal system but then for publication dumb the results down to ordinal.
Seriously, why is that?