ORACLE

Mitt Romney’s parable

By Doug Samuelson

“There!” the computer technician exclaimed triumphantly as he stood up from the chair, inviting the OR/MS analyst to resume her place at the desk. “That should fix your problem. Try it now.” “Thanks,” the analyst replied as, indeed, her screen came back to life. “At least it seems I didn’t cause the problem, which is reassuring. I have to admit, being good at using these things, especially for complicated model-based analyses, doesn’t mean you know how to fix them when they crash. Is it just me, or are they getting harder to maintain?”

The computer tech replied, “It’s not just you. As we want computers to do more and more, and do it with more resources, more data and higher speed, they get more difficult to design and maintain, and harder to debug, too. And then there’s the user-arrogance factor.”

“I know about the complexity effect,” the analyst agreed. “One of my colleagues pointed out a few years ago that the more closely a computer model of social or organizational systems approaches the complexity of real life, the harder it is to distinguish genuine rare events from deficiencies in the programming. I hadn’t thought about it before, but I suppose something like that idea applies to debugging operating systems, too – the more things we expect them to do, the harder it is to think up, let alone test, all the odd conditions that could cause a crash. But what’s the user-arrogance factor?”

The tech gave her a sly grin. “I don’t know your politics,” he said, “but we got a really good example about a month ago from the Romney presidential campaign.”

“What do you mean?” the analyst asked, intrigued. “Actually, I kept up on a lot of what happened in that campaign – and got about as sick and tired of all the blather and negative ads as most other people. I followed the issues and noticed how the candidates’ positions and public appearances seemed to be heavily influenced by some pretty high-powered focused polling. And I thought some of the top people around both candidates, and a lot of the media types, seemed awfully full of themselves.

“There’s even a story from 2004 about how all this attention got focused on the counting problems in Ohio, but Ohio went for Bush by 118,000 votes – so even if there was some skullduggery, which I think there actually was, it was hard to find enough to change the outcome. But meanwhile, three other states, Colorado, Nevada and New Mexico, with a total of 19 electoral votes, went razor-thin for Bush. Those states’ Democratic chairs were screaming at their national people that one more TV buy on the right issues would carry those states. Kerry finished with a $25 million surplus, so the money was there – but he just never made the decision. But you mean something different, don’t you?”

“Yes, indeed,” the tech affirmed. “I have a friend who worked on the Romney campaign’s targeting software, so I got more than came out in the media – and there have been some media stories, as well. They had this new, high-powered computer program called ORCA to run their Election Day get-out-the-vote volunteer effort. As you know, both parties make a massive effort to identify who’s likely to vote for them and make sure those people get to the polls. That includes door-to-door canvassing and phone calls to remind people, offering rides to people without cars, and having pollwatchers at lots of precincts making sure the other side isn’t cheating somehow.”

“Right, I know,” the analyst acknowledged.

“Well,” the tech recounted, “they had this new, improved approach, tied to their issue polling, and then that produced detailed directions for where volunteers were supposed to go and what they were supposed to do there. But these designers and managers were so sure of their own expertise they never beta tested the software with live people. So, surprise! On Election Day, the software crashed, and they had hundreds or even thousands of volunteers who gave up and went home for lack of direction.”

“Ironic,” the analyst laughed, “and maybe this means the system doesn’t work as badly as we feared. Kerry lost in large part because he couldn’t make a decision quickly, which was one of the more important things the other side was saying about him. Romney lost because he and his senior people thought that if top management is expert enough, the usual procedures to prevent foul-ups can be bypassed – which often fails disastrously, and that’s what the other side was saying about them. I don’t know your politics, either, but maybe the conclusion is that, in spite of all the hoopla, the voters tend to get it right!”

“Maybe,” the tech shrugged. “I was for Romney – no need to get into a big discussion about why. But the conclusion I see from this is that principles of good practice evolve for a reason, from experience, and anybody who thinks he or she or they are so smart they don’t have to follow those principles is courting trouble.”

“Yeah,” the O.R. analyst concurred. “My wiser colleagues have learned the hard way, from many failures trying to apply theory to practical problems: when experience speaks, pay attention!”

Doug Samuelson (samuelsondoug@yahoo.com) is president and chief scientist of InfoLogix, Inc., in Annandale, Va., and a senior statistical analyst/subject matter expert for Great-Circle Technologies, in Chantilly, Va., supporting various national security applications.