After 2 weeks of constant distractions ranging from security updates, through a couple of major OGRE issue reports that needed investigation, and learning how VS2005’s release affects us, I finally managed to spend some time on Kadath today. Most of that time was spent with a pencil and paper, chewing the former in between trying to draw something sensible on the latter. Basically I was planning how I take what I have now, and perform yet more automated magic to turn it into something more useful. That means taking some highly deconstructed & fragmented level geometry, required to build the initial analysis of the level, into a reconstituted mass which is both fast to render and fast to query for collisions. Basically, I’m doing to level geometry what Bernard Matthews does to chickens - mash them up into tiny chunks and then reform them in a more efficient fashion. This involves taking lots of little portals, joining them up into superportals, figuring out which ones are any good, and throwing the others away. I’m being guided by a few academic papers, but they’re written by people much smarter than me so I generally have to decipher and translate them into a form I can understand and implement.
Which brings me on to pencils. Pencils are the single greatest design tool in the world. Forget your fancy round-trip CASE tools - they’re both incredibly expensive and invite you to waste far too much time playing about with their many features. No - design is all about exploring the problem, and you can’t get any more natural and immediate (not to mention cheap) than your trusty HB. Sure, their equivalent of CTRL-Z involves brushing lots of eraser fragments onto the dining room table, but even then there’s a certain satisfyingly tactile aspect to that. You don’t have to worry about alignment, formatting - this is about exploration, so who cares if it’s messy; and it’s much faster to scribble madly on a few bits of paper, shuffle them around, and toss the ones you don’t like over your shoulder than it is to do the equivalent in a CASE tool - the implied neatness of such a tool means you inevitably spend time on detail and layout when that’s really not important at this stage.
Sure, final detailed design can benefit greatly from electronic tools, especially when there are many people on a team. But for the conceptual stages, when you need to just think about the problem in hand and not be distracted, you can’t beat a real-life blank sheet of paper. Yes, I’m a design Luddite, and proud.