Through the metasynth links in my previous post, have started listening to music generated via mathematical analysis of images. Some of it is wonky, as expected, but much of it is surprising, and surprisingly interesting. Now fascinated by this two-way generational relationship between images and music.
The most elusive goal of the perfect LJ Friends page is to display a roster of friends that represents the widest possible array of human experience, and thus to, at any given reload, be able to spontaneously generate an essentially random cross-section of human diversity in collage form.
Was thinking over breakfast today that the ability to locate an object in history is one of those uniquely human behaviors that would be extremely difficult to teach to a computer. For example, most anyone can look at a radio, an automobile, a jar, a brochure, an article of clothing, etc. and tell you with fairly good accuracy whether it was made in the 1920s or the 1960s or the 1970s or the 1990s etc. What is it that is shared by the radio, the automobile, the jar, the brochure, and the article of clothing? What marks these things as having come from a particular era? Color choices? A certain formalism of line? Fonts in fashion? This is extremely hard to pin down, and would be almost impossible to develop computer algorithms to accomplish. And yet it is an almost trivially easy task for humans.