Sanderson has devised a systematic way of recording and representing an ecosystem. He calls it the Muir web ... In the Muir web, each species of plant or animal, and each characteristic of habitat, has a full set of needs and associations, which, taken together, form a tangle of connection and dependency. . . . He then plugs them all into his customized modelling software, which places them on the map . . . The software mimics that of social-networking sites. His son, who is six years old, once asked him to explain it, and he said, "This program writes programs that tell the mapping program to make the maps that predict where all the species were."

Eighty per cent of the work is building the data sets. The glory part is turning that data into 3-D pictures.

—"The Mannahatta Project," Nick Paumgarten, The New Yorker, October 1, 2007 (full article not "yet" available online).

Universally, when computer science brushes up against a magazine subject, the journalist misses its role in the story. This very engaging story by Nick Paumgarten in the October 1, 2007 New Yorker, about an effort to create a model of Manhattan island when it was first touched by European, commits the error surprisingly.

To me, the glorious bit of the above project is that bit where someone has written a program that writes programs that guide the mapping software. This requires understanding the space of possible ecosystems, precisely enough to render it in code—which you might think is a respectable human and intellectual achievement. It calls on ecologists' knowledge, but also computer scientists' knowledge: what constitutes a representation (of an ecosystem), what parameters are needed to admit the necessary freedom, what parameters are unnecssary, and how to make such a model concise and precise.

Eric Sanderson, profiled in this article, is an ecologist working with the Wildlife Conservation Society; he has a vision of exploring Manhattan in 1609; some programmer somewhere has helped him realize this vision. I long for such a programmer to garner the same respect as the ecologist he works for.

Here's a manifesto, then.

I am interested in the "computational turn" in the history of our civilization—namely, the idea that we can explore a much wider range of imaginative objects (ecosystem models, say, or texts, or games) by creating a process that generates them. In centuries past, we were much more limited: for example, you could draw up one historical map of a place by manually placing roads and creeks in accordance with evidence. But now, we can codify our knowledge of what constitutes such a map, and then the computer becomes a toy for exploring all the possibilities. Computers are powerful not because they have an amazing amount of memory or speed, but simply because they allow us to work at a higher level, the level of defining the process by which imaginary 1609 Manhattans can be generated—rather than the lower level of making a particular guess about it.

Let some magazine reporter come along and do a piece on that.