Saturday, May 12, 2007

Ramblings about SL

Various people are commenting on Gwyneth Llewellyn's "Hotspots: Second Life's New Controversies." So, I figure I might as well join in. :)

The "Luddites"

She may have picked an unfortunate label, "Luddites," but she nailed the "Open Letter" issue. All enhancements, however far along they may be, are met with a chorus of short-sighted flamage. It's short-sighted, or perhaps I should say based in ignorance, for several reasons:

First, the law of diminishing returns, combined with Brooks' Law ("adding people to a late software project makes it later"), means that it's useless or perhaps even counterproductive to put many more people on bug fixes.

Second, big improvements in speed don't come from fiddling with code. Say you have a deck of cards, and you want it sorted. Here's one way to do it:
  1. Shuffle the deck.
  2. If it's not in order, start over again at step 1.
  3. If it is in order, whee! You're done!
This will in fact sort the deck, barring problems with your random number generator.It will also be insanely slow, taking time proportional to n! where n is the number of cards in the deck (and n! is n * (n - 1) * (n - 2) * ... * 2 * 1). If you've written a sort routine that runs like that, you can tune the code all you want; rewrite it in the tightest assembly language you can churn out... and it will still be insanely slow.

Major improvements in speed come from improvements to algorithms, not doing what you should be leaving to the compiler's optimizer. There's no point in wasting time on the old, slow version... but of course, people will continue to complain because LL isn't doing just that while work on the new version is underway.

Third, LL can't afford to not work at least part of the time on improvements. Other similar (but far more restricted--perhaps leaving creation to those 3D specialists Gwyneth goes on about--and hence easier to precalculate and optimize) systems are or will be breathing down its neck, and then we'll hear the cries that SL needs more features.

Professional 3D modelers

Ms. Llewellyn says sculpted prims have many advantages, and here again she is absolutely correct. They will allow a wide variety of shapes to be created that would otherwise require many of the current prims to even crudely approximate them, if they can be practically made from current prims at all. Eventually everyone will want items made from sculpted prims, for that improved appearance and performance improvement over what one can build from the simple geometric shapes of today's prims.

The part I disagree with is the claim that sculpted prims will create an enormous barrier to entry for building things. Only a very few "professional 3D modelers," the argument goes, have the skill and arcane knowledge required to create sculpted prims.

"popular products like Poser, for instance, sell about 150 thousand copies..." but Blender is estimated to have 250,000 users. Blender is GPL software, open source (guess those open source "tekkie" [sic] types took a break from looting and pillaging... don't they read Prokofy?), and steadily being improved in features, user interface, and documentation.

LL initially only wrote an exporter for Maya, a very expensive (Maya Unlimited 8.5 is a "mere" $7K) and proprietary package... but a few days ago a method to create sculpted prims with Blender was posted to the Web.

It's true... making sculpted prims is more involved than gluing together balls and boxes in SL. Sorry. But it's not that hard, and people are working diligently to make it easier, or at least easier to learn how to do it.

No comments: