TITLE: OOPSLA Day 1: Gabriel and Goldman on Conscientious Software AUTHOR: Eugene Wallingford DATE: October 25, 2006 2:13 PM DESC: ----- BODY: The first episode of Onward! was a tag team presentation by Ron Goldman and Dick Gabriel of their paper Conscientious Software. The paper is, of course, in the proceedings and so the ACM Digital Library, but the presentation itself was performance art. Ron gave his part of the talk in a traditional academic style, dressed as casually as most of the folks in the audience. Dick presented an outrageous bit of theater, backed by a suitable rock soundtrack, and so dressed outrageously (for him) in a suit. In the "serious" part of the talk, Ron spoke about complexity and scale, particularly in biology. His discussion of protein, DNA, and replication soon turned to how biological systems suffer damage daily to their DNA as a natural part of their lifecycle. They also manage to repair themselves, to return to equilibrium. These systems, Ron said, "pay their dues" for self-repair. Later he discussed chemotaxis in E.Coli, wherein "organisms direct their movements according to certain chemicals in their environment". This sort of coordinated but decentralized action occurs not among a bacterium's flagella but also among termites as they build a system, and elephants as they vocalize across miles of open space, and among chimpanzees as they pass emotional state via facial expression. We have a lot to learn from living things as we write our programs. What about the un-"serious" part of the talk? Dick acted out a set of vignettes under title slides such as "Thirteen on Visibility", "Continuous (Re)Design", "Requisite Variety", and "What We Can Build". His first act constructed a syllogism, drawing on the theme that the impossible is famously difficult. Perfect understanding requires abstraction, which is the ability to see the simple truth. Abstraction ignores the irrelevant. Hence abstraction requires ignorance. Therefore, perfect understanding requires ignorance. In later acts, Dick ended up sitting for a while, first listening to his own address was part of the recorded soundtrack and then carrying on a dialogue with his alter ego, which spoke in a somewhat ominous Darth Vader-ized voice in counterpoint. The alter ego espoused a standard technological view of design and modularity, of reusable components with function and interface. This left Gabriel himself to embody a view centered on organic growth and complexity -- of constant repair and construction. Dick's talk considered Levittown and the intention of "designed unpredictability", even if inconvenient, such as the placement of the bathroom far from the master bedroom. In Levittown, the 'formal cause' (see my notes on Brenda Laurel's keynote from the preceding session) lay far outside the "users", in Levitt's vision of what suburban life should be. But today Levittown is much different than designed; it is lived-in, a multi-class, multi-ethnic community that bears "the complexity of age". On requisite variety, Dick started with a list of ideas (including CLOS, Smalltalk, Self, Oaklisp, ...) disappearing one by one to the background music of Fort Minor's Where'd You Go. The centerpiece of Gabriel's part of the talk followed a slide that read
(google it)
"Unconventional Design"
He described an experiment in the artificial evolution of electronic circuits. The results were inefficient to our eyes, but they were correct and carried certain advantages we might not expect from a human-designed solution. The result was labeled "... magical ... unexpected ...". This sort system building is reminiscent of what neural networks promise in the world of AI, an ability to create (intelligent) systems without having to understand the solution at all scales of abstraction. For his parallel, Dick didn't refer to neural nets but to cities -- they are sometimes designed; their growth is usually governed; and they are built (and grow) from modules: a network of streets, subways, sewers. Dick closed his talk with a vignette whose opening slide read "Can We Get There From Here" (with a graphical flourish I can't replicate here). The coming together of Ron's and Dick's threads suggest one way to try: find inspiration in biology. -----