During my five years with Exxon, I was often in the position of seeking bids from companies that built unusual equipment at hideous expense. I sometimes found it useful to perform the reality check of asking myself, "What would it cost to do this with an army of people instead of buying this big piece of gear?" When the answer made the "human wave" look more cost-effective, I knew that it might be time to see if something was being over-designed--or if technology was being pushed beyond its current limitations.
As IT takes on new challenges in recognizing faces, understanding spoken words and even predicting peoples behavior, perhaps its useful to ask this question more often about proposed IT projects as well--and to look for the ideal synthesis between what people do better than machines and what machines (whether built from hardware or software) can do to enhance human performance of complex tasks instead of trying to replace people entirely.
For example, I saw last week that the city of Tampa has decided to abandon its exercise in using face-recognition technology to scan for wanted criminals in public places. When this project first got under way, I took part in a TV debate concerning the implications for personal privacy: At the risk of having angry e-mail messages outnumber even SoBig worms in my inbox, I welcome the idea that people might behave better in public if they thought that someone might be noticing, so I dont see a privacy problem at all. (As I said at the time, Im much more worried about the profile of my personal behavior that someone could easily assemble from my credit-card records than I am about the chance that someone will scan hours of videotape, from dozens of different cameras, to figure out where Ive been and what Ive been doing.)
I just want to know, does it work? After two years with no arrests resulting from Tampas pilot project, which was funded by technology provider Identix Inc., the cost-effectiveness argument clearly was not there. End of story.
For a different scenario of how to use public-space surveillance tools, consider this passage from Larry Niven and Jerry Pournelles 1981 novel, "Oath of Fealty"--which is set in a massive, self-contained, high-rise urban environment that takes gated communities to the ultimate extreme. A new recruit to the citys security force observes an experienced officer moving at random from one surveillance camera view to another, and asks, "Wouldnt it be better to have assigned places? Instead of jumping around?" His mentor replies, "Who can be alert just staring at one scene all the time? The math boys worked it out, how many of us, how many TV screens each, probability of trouble--over my head, but it seems to work."
In that last sentence, you may hear an echo of the "scientific management" arguments of Frederick Taylor, who sometimes spoke of workers as if they were machines: mechanisms whose operation needed to be optimized in the same way that an IT administrator tunes a server for peak performance. But Taylor also said, "The first object of any good system must be that of developing first-class men; and under systematic management the best man rises to the top more certainly and more rapidly than ever before." We can aid ourselves by carrying a PDA to ease the burden of remembering the details of our lives; we can aid our entire organizations by using operations research and related disciplines to eliminate excuses for failure.
As Niven and Pournelle famously had one of their characters say in their novel, "Think of it as evolution in action."
Research Assistant
