Exploring the Thingspace of Thingspace:
In a some conceptual discussions, I’ve found it useful to appeal to “intuitive thingspace”, a general term for a broad cluster of ideas that seem related. Risking recursive repercussions, I thought it might be interesting to explore the intuitive thingspace of thingspace.
(I’ve added “intuitive” as a prefix to make it clear I’m focusing on the relations between similar ideas that are “readily apparent”. Also, while Yudkowsky used “thingspace” to refer to the general cluster of “X-ish things” that fell into category X, I’m using it more for general intuitive clusters.)
While conceptspace looks at the ridiculous exponentiation of attributes across a huge distribution, thingspace is gesturing at the nebula of “related” terms that come to mind easily from the “seed idea”. We’re focusing on what sorts of ideas are triggered, looking for the ones that seem entangled in some way with the different attributes of the focus of our thingspace.
Basically, thingspace is the “space” of related concepts to a word/idea, the array of things that seem to share some similar traits, by means of association, analogy, metaphor, or any other connections that our brains tend to use. I visualize them a little like standard map/concept maps:
This is what the thingspace of thingspace looks sort of looks like for me:
“Conceptspace”, “ideaspace,” “mindspace”, and “meatspace” are the words that come out first. Following that, there are general thoughts about concepts, ideas, and gesturing vaguely at things in a fuzzy n-dimensional maps of thoughts. Other related ideas that appear are along the lines of models, ontology, and mental representations.
While your intuitive thingspace for the concept of “thingspace” may be different than mine, I can be fairly confident that we’re probably thinking of thoughts that are similar to some extent.
In discussions, acknowledging a shared thingspace can make it easier to gesture at the intuition you’re trying to convey. Acknowledging the existence of higher-level groupings (which center around some general ideas) could be useful for abstracting the “essence” of things and seeing connections.
I’m counting on the fact that our thingspaces tend to cluster around similar topics about minds and ideas. I’d be very surprised if your thingspace for thingspace included water balloons, for example.
Essence is a tricky thing to pin down. I’m deferring my definition, as Hofstadter does, to our intuitions:
For example, for most of us, there’s something (or group of things) that links tennis balls and ping pong balls together, for even though they’re made out of different materials, are different sizes, etc. Both are used for sports (another tricky category), are hit back and forth, end with “ball”, ping pong is also called “table tennis”, etc.
Notice that this requires some basic knowledge of human culture and proceedings. There is a lot of assumed context that makes it possible to understand these connections. If were to give a ping pong ball and a tennis ball to an alien, who had no knowledge of all this extra info, it seems likely they’d miss a lot of the “sports-ness” and “ball-ness” we associate with each object.
Thus the one crucial assumption that makes “thingspace” even a coherent notion for communication (both the concept and the concept of the concept) is my confidence that we are both running fairly similar software (i.e. the human brain). I’m counting on enough similarities between our intuitions that the ideas my brain generates will correlate well with topics that are (on some high-level) about the same as yours.
Such assumptions rest on the way we perceive things, which in turn owe much to our environment. This means that the partitions that thingspace appears to make, those high-level abstractions we appear to see, are also a result of our worldview. The simple categories we perceive can also be seen as a reflection of the resultant ways our brain handles information.
I find it really neat that the presence of absence of information can be important meta-information.
Thingspace, then, is not only a map of our related ideas, but it is also a map of our thought-mapping process. The meta-level information of which ideas are and aren’t contained within our thingspace is also a revealing part of how we process information (which is a different take of a meta-view of thingspace).