I’ve spent the last week worrying about the NBA finals, despite an aggressive apathy towards both the Golden State Warriors and the Cleveland Cavaliers. While I did enjoy Matthew Dellavedova’s brief moment in the sun, it’s not why I paid attention. I only got into the Finals this year because as part of my Grad School program, I was assigned to capture and analyze “all the tweets” with either of the #Warriors or #NBAFinals2015 hashtags. I’m no computer expert, so making this happen was a struggle. I’d get a prototype working only to find it would start throwing a weird error that I’d never heard of, so then I’d spend several hours Googling and StackOverflowing around until I figured it out. Elation was fleeting, since no sooner had I fixed the one error then another would appear. I did manage to get it successfully up and running minutes before the final game started on Monday night and promptly spent the next several hours anxiously staring at a constantly scrolling list of tweets, just waiting for something catastrophic to go wrong and blow my only chance at gathering the information I needed (that is, unless the Cavs somehow managed to win and force a Game 7, not a proposition I’d be willing to wager a grade on…). I do not recommend watching basketball this way.
A system like this is incredibly fragile, even if one builds in safeguards and error handling, as we were required to do. What happens if lightning hits the house while it’s running and blows out the internet? What about some random schmutz making its way into your hard disk? Or your computer simply freezing up for no apparent reason? The ways that computational systems can fail are almost infinite.
This got me thinking about an analogy that Hofstadter touches on periodically; the similarity of computers to the human body. These days, anyone who took 10th grade biology has an intuition about the ladder of complexity in their anatomy that moves from organ systems, to individual organs, to anatomic structures, to cells, to organelles, all the way to individual strands of DNA. What’s far less frequent, through no less important to our modern survival, is the analogous ladder of complexity in modern computing. Most of us are pretty content to think of our iPhones as excellent rectangles and leave it at that. But the idea that there’s apps written in high level languages that get translated into lower level languages and ultimately into an almost unimaginably massive number of individual ones and zeros expressed simply as high or low voltages across a piece of silicon is truly mind boggling. The number of ways that something can get screwed up somewhere along that ladder is just as large, if not larger, than all the ways that our bodies can get sick, and, in general, we have no concept of this. As Hofstader bluntly observes, “Most men don’t understand computers even to the slightest degree” (600).
Because of this lack of understanding, we generally tend to over anthropomorphize our electronic tools. I, for one, know that I have used phrases like “the computer got mad at me and died” several times just within the last week. Siri can spit out some pretty fast comebacks (try asking her about Skynet), but, as I’ve mentioned before, she’s just a bunch of really well-curated and highly-optimized natural language processing and data retrieval algorithms. Despite writing GEB over 20 years before Siri’s release, Hofstadter could easily have been describing her instead of an early NLP program called ELIZA when he notes, “this kind of program is based on a shrewd mixture of bravado and bluffing, taking advantage of people’s gullibility” (600).
Which brings me back to the NBA finals. Seeing just how difficult it is to get computers to do relatively simple tasks and scraping the surface of all the ways thing can go awry makes systems like Siri extremely impressive. The fact that they’ve reached a level of sophistication where we can even discuss them in terms of bravo and bluffing reflects years of tedious and laborious work on behalf of legions of computer engineers and mathematicians much much smarter than me. I think I’m just starting to grasp the scale of the achievement that all this work represents, and it’s an exciting feeling.
Godel, Escher, Blog 