User:Woozle/posts/others/2-Place and 1-Place Words

Link: http://lesswrong.com/lw/ro/2place_and_1place_words/

Points

• The ancient, pulp-era magazine covers that showed a bug-eyed monster carrying off a girl in a torn dress don't make any sense because they assume that the bug-eyed monster would have the same function for evaluating "sexiness" as people who bought those magazines when they were new.
  • The "sexiness" function is, rather, a two-input function, with the observer being one input.
  • EY gets very elaborate about how to formalize the relationship between people, the things they are evaluating, and their evaluations of those things.

Responses

Pulp Magazines

This always bothered me. One possible resolution (possibly not the one intended by the cover artists) is that the alien wasn't dragging the girl off because it found her to be sexy, but rather it was dragging her off for its own reasons -- and the artist chose to depict this scene rather than other {acts by the alien done for the same reason} because the artist found the girl to be sexy (and more likely to sell magazines -- or at least more likely to sell the painting to the magazine's editor).

For example, maybe the alien was collecting specimens of Earth life for examination. It carried off a small tree, a sea-slug, and a young human female. Which of those scenes would you choose to paint, if you were an ancient pulp-era magazine illustrator?

Twin Earth

The question in the "Twin Earth" thought experiment is one of those things where people try to let words do the thinking for them, and it doesn't work because the words have been deliberately subverted for the purposes of the experiment. It only matters if "water" means the "same" thing on our.Earth as on twin.Earth if the two universes interact somehow (in much the same way that "God"'s existence only matters if "God" interacts with us somehow; if "God" does not, then statements about "God"'s existence are meaningless.)

For instance, if someone (from our.Earth) might ask for water (meaning H20) and be given XYZ instead. If XYZ can serve as a functional substitute for water in most ways, then our.people might designate it "a form of water". If it turns out to be a deadly poison to our.people, then they might designate it "a deadly poison with deceptive similarities to water".

On the other hand, if a twin.person asks for water, one would assume by default that they "mean" XYZ, and only hand them H2O if one knows it to be a reasonable substitute for their current usage (preferably after asking if a substitute is ok).

Meanwhile, a scientist doing a molecular analysis of twin.water (i.e. XYZ) would say that it definitely was not water.

I guess my point is that there are lots of different meanings of -- sets of attributes implied by -- the word "water", and whether the word would apply to a subtly different substance that looks similar would depend greatly on the context.

This is one of those "rephrase the question" things -- like "if a tree falls in the woods", which EY has written about elsewhere (will add the link when I come across it again).