An Examination of The Art

Posted to the Lyst on 1/15/02, during a discussion of Descriptive Books.

I think Patrick's right on when he says:

" Wouldn't it make more sense to write the natural laws or conditions that dictate what color leaves and bark are? Things in nature are not painted from the outside rather they develop out of internal laws of biology, geology, chemistry, etc..."

The crucial passage for me is the one where Anna explains Garo-hevtee to Atrus.

She describes several "levels" of words:

1. a label for a concrete thing: tree, sand, rock.
They can be modified by labels: palm tree, fine sand.

I note that these labels are not simply adding to our knowledge of the thing. "Tall tree" would not contradict "tree". But "palm tree", while it doesn't contradict "tree", really is a substantively more precise idea than "tree". I could see that adding "palm" to a finished Descriptive Book's "trees" might be enough to shift the link to a new Age, whereas adding "tall" might not. This is a bit subjective, and probably is the dirty gray area in which Atrus had to swim in order to work on Riven.

2. labels for abstract things: love, loyalty.
These, Anna says, tend to need a great more definition to pin down, a little background info to give them context.

3. The Garo-hevtee are labels for things that are not concrete or abstract, but one level more complex than either, and it is a D'ni conceptual level which human languages don't have. Therefore, at best, we can only visualize it in the way that we can imagine what four dimensions looks like.

So the Garo-hevtee signs don't specify "palm trees" or "island"; that's level 1. Rather, they specify things like "a symbiotic relationship between plants and animals follow these patterns and tends to involve the cleaning and disposal of parasites in exhange for camouflage and protection of young."

But there are also signs, obviously, which DO specify things like "palm tree" and "island", because Gehn shows Atrus one. This is why Dar'nay argues that The Art seems to use both Garo-hevtee and normal D'ni writing. That still troubles me, but he's quite right; the books seem to imply that at least two levels of labels are being used (1 and 3).

I also like Patrick's analogy with "included files": there are certain common phrases, passages, or thematic elements which have been tested and proven to work, so they get used over and over. The D'ni seem a very conservative bunch who prefer perfection of a traditional method to innovation. So they must rely heavily on concepts thousands of years old.

This is what Gehn did. However, much like the "labels" idea, this is only "level one" of the Art.

Level two is what Atrus found through Anna's teaching and observation: one cannot simply stitch disparate phrases together from different books, however well-written. An Age is not a bunch of building blocks stacked together. Instead, it's interconnected systems-- analog, not digital, if you will. So the writing style, the flow of the book, its structure, the flavor must strive in every way to be unified, for all parts to mesh well with other parts.

For this reason, I would not be surprised to discover the last word or sign of the book could loop around equally well to fit syntactically before the first word. There shouldn't really be a beginning or end.

So how on earth could we write that?

Well, Gehn gives a simple Age with 13 signs as an example for Atrus to observe. These must rely fairly heavily on Garo-hevtee to be so compact, but the first word, Gehn says, is "island".

(Now, he may be using a too-precise interpretation of the word. Perhaps that really is a Garo-hevtee sign, and comprises not simply the label "island" but also includes structure, erosion, type of ocean currents around it, amount of rainfall, kinds of sand and vegetation, and all ideas that make up the broader concept of "island").

So apparently--IF Gehn is understanding the Art, which is a problematic if-- you start with a word that encompasses the place you mean to Describe.

Note: they seem to stick to one island, cave, or area. The rest of the world also exists and may have other islands or areas, and by extension, there's a solar system, a galaxy, and a whole universe. Apparently you can muck with these larger things. Atrus dabbles in orbits and moons now and then. But it's probably complicated enough just managing to keep unity while describing an island, and every time you add in another factor, it's like adding another juggling ball.

Once you've described general setting, you start setting out general parameters. I have a feeling these may be very broad: geology, ecology, climate. You can stop after that, probably, and the Age will fill in all the incidental details if you link. (It's fairly vital that the Age fill in non-specified objects, because otherwise, you'd have to describe every single thing in the entire universe). But obviously, this is about as vague as telling Bryce 3D "I want a terrain here". You'll have no idea what the fractal generator's going to slap in place. So the next thing is to start refining these parts, and I *think* you're allowed to get very detailed indeed, because Atrus is able to put in blue flowers and specific animals.

How are these objects described? With animals: anatomy, diet, habits... With plants: fractal structure (fan-shaped or spiral leaves, spreading root or taproot), water conservation, metabolism...

There's one other thing, though, where you can get very specific. Gehn says, "For a Linking Book to work it must also include an accurate description of the place one wishes to link to on that particular Age, which is recorded by writing a special D'ni symbol, a Garohevtee."

I don't think he means the Garohevtee describes the entire place all in one symbol; that would be level 1 of the Labels theory: concrete things. Rather, I suspect it's a combination of level 1 labels: the things in the area, and then a Garo-hevtee labelling these labels as the linking point.

If Linking Books can be that specific, describing the features of one location, then Descriptive Books surely can be as well. And we have evidence for that too. Atrus copies the Cleft very accurately for his first age. He can't put in the structures, but he can copy the shape and composition of the rocks.

This starts getting tricky. In general, Descriptive Books must refer to factors, parameters, trends in the Age, rather than naming fixed points, because forests evolve, mountains erode, ecosystems change. So how can you get away with describing the shape of rocks? The Age shouldn't collapse just because some kid carved their initials into a cliff!

Dar'nay again gave me a hint: The Description presents a default state at the moment of Linking, but does not preclude natural change from that state, and may optionally suggest ways in which change occurs (i.e. "glacier here, which tends to move and melt at an average rate of X").

This would explain why the Linking Window is blank until the moment you Link, and then the Age is "fixed".

I believe that specifying place and land forms in this fashion would probably fall under the category of "initial description of setting".

I am seeing the structure of the book as something like collapsing menus where the sub-menus are optional and recursively more specific:

(Let me take the Big Island of Hawaii)

  • General setting: Island
    • volcanically active
      • mostly pahoehoe lava
      • shield volcanoes
        • two volcanoes, northwest to southeast; southeast one has a very active sub-volcano on flank *
        • Pu'u O'o vent: about 800 meters high, currently erupting * continuously
        • active lava flow going into the ocean at x point on the coast*
  • Climate: tropical
    • Season length, rainfall, etc....

The (*) items are things that are specified as the "initial state": what's there on first Link, but which obviously are going to change over time.

As soon as I started trying to write it out, I run into the problem that English can *only* describe specific objects, things which indeed are going to change over time. That must be where the Garo-hevtee comes in. So the above "sample Description" is equivalent to a 2-dimensional representation of a 3-dimensional space.


I just proposed a very organized way of writing an Age. In other words, stylistic conventions. These include the order in which things like climate and topology are specified, and what sorts of things go where.

Other D'ni laws seem to be: no artificial structures, no meddling with the book after first Link.

They *say* these are laws. But the games of Myst seem to be about proving those laws are actually conventions, passed down by long tradition, which help keep the writer from producing unstable or unpredictable ages.

Gehn and Atrus edit Ages after they've already been visited. This sometimes helps (Atrus forestalls Riven's decay), sometimes breaks the link (when Gehn mucks with Age 37 the second time), and sometimes introduces factors which, when interacting with others, can have unpredictable and disastrous results (making the sea warm changes the climate, diminishes rainfall, and causes the sea level on Age 37 to drop for some reason).

It occurs to me that in addition to Description which specifies the initial state of the Age upon arrival--probably, like the "linking point" in a Linking Book, a description labelled with a specific Garo-hevtee character-- you might be able to specify a state at some later point in the Age's development. And *this* could be added in as an addition later, like the changes mentioned above. It's the only possible explanation for what Anna and Catherine did to Riven. Furthermore, in addition to being able to specify a point using physical landmarks--as one would do for the linking point--Catherine and Anna seem to have been able to use a point in time instead.

I suspect, again, that's breaking standard D'ni convention and getting into some very tricky stuff.

(Side note on Time Travel: since I posit the Descriptive Book is naming a default state at the time of "first link", you can't really travel to a different time, even by specifying a different time point in the Linking Books. I think they'd just fizzle. You control the state of arrival, but the Tree of Possibility chooses a point in time which bests fits the description of that state. If you tried to specify "same spot, different time" somehow, you'd simply link to a parallel Age, not the same one, much like the parallel Age 37. My pet theory is that this is what Atrus was doing with the planetarium on MYST: using constellations as "fixed points in time" to try to link to different times in an Age. But he could only have linked to different parallel Ages using this method.)

Back to laws of the Art. The D'ni forbid artificial structures in Description. Obviously Catherine thumbed her nose at that rule with the daggers--probably without a plan; she seems to write like a poet and just put in what the muse tells her to. Atrus, baffled by the fact that his wife was able to do that, tried bending this rule himself and discovered how very unpredictable the results can be. Apparently one can describe a structure, but the Age plops it into the setting at random, and much like Bryce, seems to drop it into the landscape and partially embed it.

Finally, the flowchart--the structure of the Description--the collapsing menus-- seem to be optional. If you don't follow a structured pattern like that, you're liable to have chaos. As far as Atrus knew, you *had* to follow a structure. But Catherine, again, shows that you don't have to--it's just a lot easier to keep track of. She managed to make a unified whole in her Torus-world by stream of consciousness, poetry, and nonlinear thinking. Atrus thought her Description broke the laws of the Art, but in fact, it was only the conventions.


That's more than enough, and I apologize, but I've never really tried to put together all my thoughts on Descriptive Books before to try to come up with a Theory of the Art.

And much like Catherine, this was all stream of consciousness, unplanned, and unedited, so you must judge for yourself whether it makes sense or has any structure. ;)

I wrote this after reading Dar'nay's discussion of the art, so I am heavily indebted to him for sorting some things out and presenting the relevent textual evidence from the novels in an organized list.