This is a Glass Bead Game post, but it might be applicable to the endless debate over "sandbox" versus "railroad" RPG adventures...
A key problem that I've faced when trying to design some kind of real-world variant on Hesse's mythical GBG is the issue of the symbolic language. Hesse's game wasn't supposed to be a free-for-all conversation, where people just talk about comparing and contrasting ideas across various fields. It's supposed to utilize a kind of "universal syntax" that formally allows deep and exact comparisons to be made -- and archived -- between far-flung ideas. But developing such a language is both (1) more than a life's work!, and (2) a pretty high barrier to entry for people to "just" play a game.
Thus, different game designers, eager to have something ready to play in our lifetimes, have taken many different paths with regards to the formal structure of their games. Let's look at two interesting extremes...
1. William Horden's Intrachange. This GBG variant takes advantage of some deep similarities between the mathematics of the 64-square chess board and the set of 64 hexagrams of the Chinese I Ching. Each move of a possible chess game can be made to correspond with a specific pattern of the archetypal "book of changes." Even the simplest of games has multiple co-existing levels of formal interpretation. I think that delving deeply into Horden's symbolic correspondences would very likely be rewarded with new insights -- but is the effort worth it?
2. Terence MacNamee's Ludus Sollemnis. (No website, but Charles Cameron has archived at least one game here.) Here, a game is presented like a liturgical play. There's a straightforward series of statements made in the game, and each of them can be illustrated by showing a work of art, reciting a poem or other text, playing a song, and so on. People viewing or reading the game are given "glosses" on each item (i.e., instructions for meditating on the item, or links for further research). All of the ways that the items can be compared and contrasted with one another are implied by the dramatic presentation, not stated explicitly.
Two very different approaches! I wonder if these differences have their roots in the famous divide between the Two Cultures of the hard sciences vs. the soft humanities.
Subjectively, I find myself liking MacNamee's approach more, since I often find it hard to follow the "golden thread" through Horden's structures and lists. However, I think the Ludus Sollemnis is still not a perfect fit to Hesse's ideal GBG either... for the same reason that I found Charles Cameron's Hipbone games a bit unsatisfying: still not enough formal structure!
Now, the cliched approach, when presented with two straw-man extremes, is to just blithely say "We need to find a middle ground!" But the devil is in the how! Just like poetry and musical composition, one needs just the right amount of structural constraint and rules, not too much and not too little. One needs to know those rules inside and out (and to know when to break them!) in order to build insightful and satisfying works. A well-designed GBG variant just can't get around this particular Goldilocks problem! :-)