Flutterby™! : On obsolescence

Wrote this for the Metafilter entry that linked to Tim Kreider's musings on Phil Tippet's Mad God, and stop-motion animation and a craft becoming obsolete.

I now want to watch Mad God, but I'm stuck on the obsolescence part of this.

Several decades ago, I was obsessed with 3d graphics. I was writing renderers, and figuring out ways to put pixels on screens faster. That obsession led to moving to California to join Pixar's Interactive Group, and when the Interactive Group was dissolved I was given the option of becoming a "Technical Director", or going into the Graphics R&D group.

At the time I saw being a TD as being an interface by which the director used the computer. That dramatically under-valued the role that the TD brings to the process, but is also not entirely wrong. I went with going into Graphics R&D, and after a little while there realized that wasn't my place either.

Anyway, here I am decades later, caught up in a world where everything is layers of abstraction. I'm no longer obsessively counting cycles, figuring out which machine language idioms splat those pixels faster. Hell, I'm no longer trying to remove unnecessary floating point operations from geometry transforms. I have an intuitive feel for how 4x4 matrices work that many people I've run into in computer graphics don't have, and yet here I am with those skills completely unused in my current career path.

I'm about to just trash a whole bunch of books with various really cool math and techniques because I'm never gonna write my own polygon rasterizer again, never gonna worry about finding the surface of an implicit surface.

And I miss that. I spend a lot of time these days trying to figure out WTF the people developing AppKit or JavaScript idioms were thinking, a lot of which feels like unnecessary layers of abstraction on top of layers of....

And I think I can draw an analogy straight from "this brush stroke goes right here" vs "let's change the input sentence a little bit and see what that gives us" to my own path of worrying whether LEA was faster than ADD and MUL vs how the framework that some coworker pulled into the project is abstracting out threading in a way that's causing us a deadlock.

I feel a lot of that "in love with a dying art", even if that art was assembly language. I fill the part of my head that used to be obsessively staying up late at night pondering state machines with square dance choreography, of all things.

And I'm reading about how JIT languages are edging in on compiled languages for performance, and thinking about what skills I should be developing for my next act. What's the art I'm trying to express, rather than the craft that gives me joy?

Which loops back around to computer animation and ML generated images and even the local steam power enthusiasts group, and pondering how we fall in love with the craft.