Words Made of Books 2
Published by Jeremy Douglass April 19th, 2005 in search, ASCII.
After reading Mark’s previous post on Amaztype, I played with it a bit more and had some additional observations on how Amaztype works, some of the effects of their design decisions, and how any of this relates to ASCII Art and Search Engine Art.
The Idea of Books
The individual books are clickable, but this isn’t generally something that you can act on in an informed way - you can’t make out most individual books in a word of more than four letters with a browser set to say 800 pixels wide. Amaztype words aren’t about the books so much as the IDEA of the books. Most of the tiles could be filled in with a color gumbo and I’d never know the difference - rather I understand and believe that they are drawn from Amazon’s public metadata API, and that knowledge makes me feel that the word being generated before me is full of meaning.
Two parts of the Amaztype aesthetic seem to run counter to each other in an interesting way - the random angling of the tiles (which makes them resemble physical objects tossed in a pile) and the random resizing of the tiles (which completely breaks the illusion that each image is a physical book by varying size so much both between books and even across the same book image tile).
Of course, there is a very good design reason for random resizing. The algorithm used to randomly populate the space inside the letters would chunk badly on curves and render many letters unrecognizable unless small tiles were used to fill them in (or unless larger shapes were allowed to fill against the curves for a very long time - which they can’t, see below). Rather than using uniformly small tiles, Amaztype varies the tile size randomly, or even filling a uniform tiled grid as one would in a photo-mosaic, Amaztype emphasizes the chaotic emergence of the letters through the random fill of Words Made of Scalable Book-Images.
Decomposing
Chaos is also present in another aspect of Amaztype - continuous loading, which after a certain number of books begins to create an odd stripping-away effect from the bottom layer of the image. The letters reach a point of maximum fill, and then begin to decompose. This is because there are a limited number of tiles generated, and once this limit is reached an old tile is removed from the bottom layer for each new one that is added. Previously filled areas of the text begin to look ragged around the edges as the total number of tiles continue an endless reshuffle around the bounding letter-space, and I found myself (in taking these screen-captures here) trying to waiting for the perfect moment when chance favored an even distribution and the letters were at their most clear.
Not Found
Amaztype has an interesting not-a-bug-but-a-feature - the error behavior when it cannot find any books which match your search. Rather than an error message, the engine actually passes the text “Not Found” to the generator, which then fills the words “Not Found” with the images of books that contain (naturally) those words. It would be interesting to see this kind of behavior in another search system with a richer vocabulary of potential messages - “Saved” “Deleted” etc…
ASCII and Search Engine Art
Amaztype uses collage composition in a way utterly unlike ASCII Art or photo-mosaics, both of which are generally characterized primarily by unit regularity (a grid, fixed distance fonts, etc.). Like other search engine projects built off public metadata from Amazon, Google, etc, part of the interest in Amaztype is not what it communicates, but what it doesn’t - the way that a mass of information looks when you pull back and unfocus your eyes….
Is each title only present once on any given search image?
To answer your question Mark, no. I’ve seen the same book over and over again.
I entered ‘Narrative’ and was delighted to see books that I’ve got on my shelf. I felt like the whole history of Narratology was being slowly dotted to form a field, to form a word. I too waited for a good point to grab a shot and even recorded a movie file of Narrative emerging. I waited in front of that screen, waited for the end-point but it never came. Then I set it to record and went out for a few hours. I returned to see the same word filling and decomposing and a file too big to save. Of course Narrative has no ending! What was I thinking?!
But I thought having the word ‘Narrative’ or some other relevant term slowly building behind me on a big screen as I give a paper would be a good image. I’ll definitely do that sometime.
I’m suddenly realizing that I knew the answer to this question (see:Remediation), but then I’m wondering if you or Jeremy could redescribe the process by which the word begins to decompose.
Unlike Jeremy I didn’t see the bottom layer being removed but new specks (tiny pics of book covers) being added to the fringes of the letter shape. This had the effect, for me of a filling of space and then suddenly a reduction of the space filled. In one moment the edges of the letter was bursting at the seams, about to fill an outer speck, and then another speck is added even further away, thus making the space already filled seem smaller. It is like play with perspective I guess.
I’ve been watching the creation of the words again and have noticed this time that the books do disappear as well as appear. So the decomposition is an effect on the eye and an actual taking away.