20 Questions

How many names are in your world?

In the game of 20 questions, one person thinks of a thing, and the other person must guess it by asking no more than 20 yes-no questions.

The fact that this game is interesting—as opposed to trivial, or impossible— suggests that humans have names for 220, or one million, things in their world. If humans only had names for one thousand things, then ten questions would suffice to specify any one of them; if humans had names for one billion things, then 30 questions would be necessary.

How do you slice your world?

By watching how people play 20 questions, we can discover not just the size of their world, but also how they structure it.

By tradition, a game of 20 questions begins with the fixed ternary question

Traditions like this begin because experienced players find that they always start the game with the same one or two questions, reflecting the two or three broadest conceptual categories in their model of the world. The fact that people begin the game by asking "animal, vegetable, or mineral" suggests that they organize their world like this

                 everything
                    | 
           +--------+----------+
           |        |          |
        animals   plants   non-living
                           things

Twenty-first century slices

The game of 20 questions dates back centuries, and the question "animal, vegetable, or mineral" is more than 100 years old. As such, it reflects the world view that people used to have.

When I play 20 questions, I don't find the "animal, vegetable, or mineral" classification so useful. Instead, I begin by asking

which suggests that I organize my world as

                everything
                    |
             +------+------+
             |             |
	  man-made      natural
                           |
		      +----+----+
                      |         |
	            living  non-living

This probably reflects the ascendency of technology in our lives over the last 100 years.


Notes

centuries
conceivably, to the origin of spoken language

Steven W. McDougall / resume / swmcd@world.std.com / 2001 Aug 09