Last week, I pointed to an algorithm commissioned by Congressional Quarterly for predicting the outcome of Congressional elections based on a number of factors, including the occurrence of the candidates' names and certain keywords on Google, the performance of presidential candidates in the 2004 election and others. Search Insider today comments on the algorithm's performance (3-of-4 correct), saying the consumer search is not the ideal way to predict an election.
We agree, but we differ with Search Insider on the point that the algorithm was misdirected. Search Insider says "[The algorithm’ author]'s mixed record ... speak(s) to the unpredictability of elections. Consumer search behavior was never designed to be an indicator of election results." In fact, the algorithm made no use of consumer behavior. It merely counted the number of hits on a phrase, which is not a search behavior but a characteristic of the data set that is the Web.
Moreover, the exactness of the calls where the algorithm was correct were remarkable. Not only did it call the Virginia Senate race as a squeaker won by Democrat Jim Webb, it correctly anticipated the relatively handy victory by New Jersey Democrat Robert Menendez over an opponent who was thought to be narrowing the race. It also called the Montana Senate race correctly.
The algorithm was wrong on one election, the Nebraska Senate race, won handily by Democrat Ben Nelson. But, as we pointed out last week, we all knew that was a flaw in the algorithm at the time. CQ said it was wrong and it turned out to be so. The algorithm needed some tweaking.
But, let's be clear, it was not an algorithm for winning an election, only for predicting its outcome. Winning an election requires a plan and a lot of flexibility.
To illustrate this distinction in a blog setting, let's look at the recent discussion of AOL vice chairman Ted Leonsis' blogging strategy, which involved blending his name with those of celebrities and sports teams to raise his Google Page Rank:
First, he started blogging. In January he launched "Ted's Take," his personal blog, posting daily snippets of his life that offered glimpses into the world of one of the area's richest businessmen. For example, in an entry from Nov. 6, Leonsis shares his busy schedule: He gave a speech at the University of Maryland's business school. He met singer Lionel Richie, who gave a private impromptu performance. And he watched the Caps lose to Atlanta from his box at the Verizon Center, where he played host to the rock band Barenaked Ladies.
He planned his rise in the search rankings. But he did not, as Nick Carr suggests, control the conversation. He applied a search engine optimization strategy and measured the results (by Googling himself), building on successes while dropping efforts that weren't paying off. As Leonsis told The Washington Post how he achieved his goal and sent messages to friends saying, "My job is done!" Not even close, if we are talking about doing more than gaining PageRank.
A conversation, unlike a climb to the top of Google's results, is never over. Leonsis applied a successful strategy for raising his blog to the summit of Google, but he did not control the conversation along the way. He largely ignored the conversation, focusing on his keyword blending to borrow some juice from celebrities.
If someone had decided that Leonsis had been abusing his own access to celebrity or that he was just a jerk for talking about what a cool life he had, this story could have turned a very different direction. For instance, what if a satirical site had appeared that mocked Leonsis' apparent, though planned, obsession with blogging about encounters with fame?
Control is illusory or, at least, only appears factual in retrospect. A blogging campaign that sets out to control a conversation will end on the rocks of despair if it doesn't have a very good compass and maps of the route it is going to take. And because conversation develops in real time, those maps are going to change, so there must be an attentive eye on feedback mechanisms.
Where Carr is absolutely correct is his comment that "It's about seeing the structure in the mirage of messiness," referring to Cluetrain co-author David Weinberger's comment that "Hyperlinks subvert hierarchy.... Hyperlinks have no symmetry, no plan. They are messy."
Bringing it back to the election prediction commentary, Search Insider writes:
Most consumer decisions aren’t so simple, as consumers generally have more options, more uncertainty, and no set time constraints where they have to act on one day only. For that reason, consumer search behavior’s research value should gain every marketer’s vote of confidence.
Links are an evolving structure. Where there is structure, there is leverage. And where there is leverage, there is the opportunity to change human systems.
The keystone of success social engineering is being able to test your original plan against the results you receive, then adapt and move on toward your goal. As Doc Searls, another of the Cluetrainers put it, "[Leonsis] is being both Machiavelli and Cluetrain compliant."
Human systems are open to many forms of manipulation. In conversation, however, there is always the chance a strategy will be met and beaten insensible. You can't exert control, but you can be sensitive to the results you are getting and adjust your strategy and tactics to stay on track to your goal. So we need our fingers on the pulse of the network during the engagement and not just after if we're going to see our plans through.
