A MILLION MONKEYS

A Blog about AI, UI and HI

January 2007

We've been doing a lot of building and coding in Second Life over the past few months. And it's occurred to me more than once (read: "hourly") that one of the great things about the SL environment is the range of opportunities it creates for fertile improvisation. The LSL script language and building tools are rich and (somewhat) well-structured, and -- like all developer toolkits -- clearly intended to promote and facilitate certain kinds of approaches to problem-solving. But they haven't yet been surrounded by an institutional envelope that dictates thinking about best practice. Too many diverse minds are working simultaneously on too broad a set of problems.

Instead, you have a lot of very smart people in there, hacking -- coming up with wild ideas for builds, then creating functionality any way that works. Case in point: the other day, I was flying through a seashore sim and saw, out of the corner of my eye, a couple-hundred meters away, a very credible seagull, soaring through the air in beautiful spirals.

I think to myself: Oh, neat! Somebody used sine and cosine functions and conic sections to drive that bird-object-set through the air in linked expanding and contracting spirals. But when I flew in to take a closer look, I realized this wasn't the case at all. In the SL environment, there's a way of setting an object so it rotates on its axis -- sort of a one-shot, set-and-forget throwaway property that makes it easier to fabricate helicopter propellors and rotating waterbeds, right? But in this case, some anonymous genius had used it to create a kind of Calder mobile stack of invisible rotating arms, one atop the other, all with different lengths and rates, and mounted the seagull at the end of the topmost arm. The result was a beautiful, counterintuitively complex effect obtained, as it were, by analog means -- computationally altogether more efficient and sim-friendly than my gut-level, "throw numbers at the problem" approach would have been.

A far more jaw-dropping example of this kind of felicitous improvisation is found in the SL sim "Svarga," (slURL), a huge, functioning ecosystem. Designed over the course of a year by the artist Leukosargas Svarog, and embellished with beautifully-designed skyways, observation platforms, libraries and education centers, the Svarga ecosystem ties together quasi-independent artificial lifeforms in a living mesh: plants of many types and diverse characteristics, bees and birds of several species that pollinate them and spread seeds, and some clever enhancements to the SL weather system that generate a random water-cycle across the terrain.

The result is a lush, fully-functional, organic environment -- because the flora and fauna are reproducing and cycling independent of direct human intervention, they've arrayed themselves in forms of great and mysterious balance and complexity, which change constantly -- spend a few hours in Svarga, and you actually can "watch the plants grow." And the whole hypnotic, gorgeous deal is apparently (i.e., from what we've been able to judge by peering inside a few plants and birds) generated using code little more complex than the four-rule Conway Life simulation, implemented by distributed scripts colluding with one or more cleverly written central controllers. Though one detects that the real "intelligence" of Svarga isn't so much in the code that ostensibly drives it, but in the thousands upon thousands of unvoiced aesthetic/technical decisions made in creating the physical forms of organisms that inhabit it. And this, of course, whether it was created in this place by the artist's instict or meticulous intent (jaw-droppingly impressive in either case) is how nature really works, right?

Posted by John Jainschigg at 12:49 PM  Permalink |