“My Secret Hideout” is a new iPad application from Andrew Plotkin, which bills itself as interactive art or poetry or a toy. Technically it’s a bit easier to describe: it’s a text generator manipulated by an intentional cryptic and evocative graphical UI. You drag and drop new nodes onto a tree, and the text on the lefthand side changes in response to what you’ve just done. The structure you build maps to a random seed that determines the textual output. Move nodes, and content changes. What it creates is a description of a location or a string of linked locations: IF-like room descriptions, only without the ability to manipulate the objects directly. Consequently it shares some of IF’s appeal — the visionary access to a place of someone else’s imagination. There are a lot of zarfian ideas here. The secret hideout you produce will likely feel like a Myst-like place, a dwelling for mystical inventors, combining machines and instrumentation with natural materials.
For instance, the tree image here produced the following description:
My secret hideout is a ring of beautifully-ornamented cubbies hanging in a pine stand. A doorway, engraved with territorial diagrams, opens out to a core room, overgrown with ivy.
The hideout is powered by a miniature steam boiler, rattling cheerily on a side platform, in a small cabinet.
An observatory is on the left side. A high-powered telescope is set up by a smoky grey beanbag chair. An antique mechanical clock stands on a pile of blocks. A bookshelf of astronomy reference books stands to the side.
This isn’t in any meaningful sense a game: there aren’t any goals, scores, win/loss states, etc., and it’s hard even to see how one might project such things onto the structure.
It’s also not a story. I’ve often argued for the power of setting as a story-telling mechanism, and the significance of objects as conveyors of narrative, My Secret Hideout doesn’t entirely respond to that kind of treatment. The descriptions are too flexible, the whole output too mutable and dreamlike. While it’s possible to make up theories about why the narrator built this particular hideout and what it all means, there’s too little control over content — or opportunity to select and label favorite content yourself — to encourage that mode of thinking for very long (I found, anyway).
To the extent it is a toy, the toy-nature is about figuring out how it does what it does and how much agency you can reasonably exercise over the output. As you may want to work that out for yourself, the rest of the discussion might be considered moderately spoilery, hence the cut.
And then I’ll talk about what I think it is, and to what extent it’s good at being that thing.
Here is my experience with how it works: adding a node to the trunk elaborates the first room of the hideout. Moving that node produces radical changes in the appearance of that room, but it pretty much always makes some reference to a central room or sleeping area. Adding more nodes in a line produces more complexity in the initial room description, though with diminishing returns, so a linear string of many nodes is likely to be not-that-different from a string with one node fewer.
Adding a new branch to the tree introduces a new location, presented in a new paragraph. Tweaking the nodes on this branch only tweak the new location. This makes it possible to feel you’re redecorating with some intention, because you move nodes around on a strand until you get a room you like and then you can move on by adding a new branch. This all feels pleasingly organic: if you could grow text like a bonsai, trimming here and expanding there, this is what it would be like. And I enjoy the aesthetic of trying to make my tree look good as well as trying to evoke text I like.
One frustration to the text-gardening metaphor is that if one of the later branches becomes longer than the original string of nodes, that branch takes over as the randomizing seed for the first room description. That means that all the rooms in the hideout suddenly change because they’ve all swapped seeds. There is an undo button, so if you repent what you’ve just done you can take it back — this isn’t instant death to your whole project. Nonetheless, I found this a bit frustrating; it feels like a frequent, strong reminder that you’re not really the boss of the text. So: yes, a toy, but an occasionally startling, disobedient one.
Interactive art, then? I think you could make at least the argument that it is, or is trying to be, a magic crayon in the Chaim Gingold sense: it gives you moderate controls and produces colorful results that are more polished than the average person could produce alone. But I think it doesn’t get all the way there, because (and how is this for a subjective remark?) I wish the output were more beautiful. I like the structure and body shape of the tree above, but there are a bunch of ways I’d like to tweak it to be more appealing.
Node shape is one of my bigger gripes. It’s not at all clear to me how (or indeed whether) your choice of node shape affects the output. I was able to swap nodes from one shape to another with only minor changes in the text, and I’m not sure whether those changes related to node shape or just to the fact that I’d placed the new node in a slightly different place from the previous one. (I was vaguely hoping that choice of node might determine content to some degree — whether that was as general as “this node type produces more built objects, that node produces woven or textile objects” or as vague as “things made with this node are described as smaller” — though I can also see that having too rigorous a system of this kind would constrain the output more heavily than one would want.)
So maybe the node choices are mostly there for aesthetic purposes, but if so, I’m not crazy about the shapes. They’re not bad shapes in the abstract, but I wish they looked better in combination (to my taste, trees look best when composed only of one node type), and that they came down more on one side or the other of the organic/machined divide. Given the content of the game, with its steampunk boilers and ratcheting instruments, it would be cool and reasonable to have a few gearlike/stiff connectors as well as fluid tree-and-branch-like ones. (The fact that I’m insistently calling these nodes rather than “leaves” — as the app does — is also significant; a couple of them have leaf shapes, but they don’t act like leaves. They act like branches.)
I also found myself wishing for changes to the sizing and leaf coloration. One of the nifty things about a real tree is the diminishing of size out at the edges — the branches get thinner. What if the outermost nodes here were smaller than the ones nearer the trunk? Coloration, meanwhile, shows how recently you interacted with a given branch or node, which is moderately interesting information, but not really necessary to have, and it again produces a kind of particolored blotchiness that I didn’t care for, as though a couple of limbs of my tree had gone dead after I’d been working on it for a long time.
Is it interactive poetry? I wouldn’t call it that. The text is good. It doesn’t have the stiffness and predictability of the average generated prose. The sentence structures vary; sometimes there are adjectives or dependent clauses, sometimes not. Subtle changes in node placement create equally subtle variations in sentence structure. Connective tissue within a paragraph usually works pretty well, though leads in to new paragraphs (typically introducing new sub-locations) are a little vaguer, presumably because they can make fewer assumptions about what has gone before.
So I don’t have a clear criterion on which to reject the label of interactive poetry. Some possible reasons: it is structured like prose; it doesn’t feel as thematically or conceptually rich as I expect poetry to be; the interaction isn’t an interaction with the ideas but is entirely at a formal level, because you can only encourage existing trends, and can do so only in the sense of making more or different sentences about the same idea.
Contrast Jason Dyer’s Renga in Four Parts: a much less procedural piece, certainly, and with less effective agency, but one that encourages the player to respond to the ideas associatively. For that matter, Andrew Plotkin’s own Space Under the Window demands engagement with the content rather than the shape, even if just by making the player indicate what s/he’s interested in, and hints at feelings and characters beyond itself. By contrast, My Secret Hideout is primarily about itself, particularly the pleasures of esoteric technical creation: many of its core images are about the creative spaces in the hideout, the owner’s tools and toys, inventions and puzzles. Does it have anything much to say about this, other than “technical coolness is cool”? I’m not sure.
So I enjoy My Secret Hideout, and I’m sure I’ll get it out again in the future. Turns out, technical coolness is cool. But I found myself wishing that the creative aspect of the experience reached deeper; that there were more room for mastery; that the discoveries you can make were more varied in type and significance.