We don’t just live in online filter bubbles.  Increasingly, Americans are surrounded by people who have the same tastes, lifestyles, and political opinions as they do.

Bill Bishop and Robert Cushing call this “The Big Sort.”  In 1976, just 26.8% of Americans lived in a “Landslide County” where one presidential candidate won the county by more than 20%.   By 2004, however, 48.3% of Americans lived in Landslide Counties.

1976

(landslide counties are shaded in)

2004

This is the context for the filter bubble.  The algorithmic personalization of our information diets is occurring at a time when Americans are much less likely to interact with people who have different opinions from them than they were in the past.

Right now, personalization algorithms amplify this trend.  But with some creativity, courage,  and intentionality, it is possible that one day soon they could be a healthy counterweight.

Can we give Ashlee Vance over at Bloomberg early props for coining the term the “Wants”?

Just like their Quant brethren over on Wall Street, the Wants depend on algorithms and massive computing power to churn data for economic gain – but instead of gaming the market, their gaming you to click on ads.

At social networking companies, Wants may sit among the computer scientists and engineers, but theirs is the central mission: to poke around in data, hunt for trends, and figure out formulas that will put the right ad in front of the right person. Wants gauge the personality types of customers, measure their desire for certain products, and discern what will motivate people to act on ads. “The most coveted employee in Silicon Valley today is not a software engineer. It is a mathematician,” says Kelman, the Redfin CEO. “The mathematicians are trying to tickle your fancy long enough to see one more ad.”

Vance profiles one Quant-turned-Want in particular, Jeff Hammerbacher, the first Facebook employee charged with making sense of the social site’s vast troves of personal data. Perhaps the industry’s first Want apostate, Hammerbacher jumped ship in 2008 after sensing he was part of a new tech bubble.

Unlike other inflated markets, which usually leave infrastructure hardware and software in their wake, Hammerbacher feared this bubble would just leave a legacy of wasted minds.

“The best minds of my generation are thinking about how to make people click ads,” he says. “That sucks.”

Luckily, Hammerbacher’s career trajectory belies his concern. The data guru has founded a new venture that aims to give companies even greater processing power. But instead of using data to figure out how to get people to spend more time on Farmville, it’ll help researchers, say, decode cancer genes and work more quickly toward a cure.

We knew this day would come, if only because we saw Minority Report: A new NYC-based start-up, Immersive Labs, makes outdoor ads that scan your face as you approach and guess your age and gender. Combining this info with data about the weather and nearby social media activity, Immersive’s billboards then serve up the ad most likely to get your attention. They gauge success by measuring how long your eyes linger on the chosen ad. And the system gets smarter with each new passer-by.

While these ads raise obvious privacy concerns, Immersive Labs CEO Jason Sosa told the Huffington Post that they don’t save face-scans or collect any personally identifiable information. Still, one wonders how far companies like Immersive are going to take this. Right now, the ads just scan for age and gender, but might they one day add, say, race or body-mass index to that list? If they can pick up Foursquare and Twitter data, will they soon personalize ads based on data pulled from your mobile phone as you walk by?

Of course, targeted ads are nothing new — we’re used to seeing them pretty much all over the Internet. But Immersive’s ads raise the question: is there something about the very public nature of a billboard that sets it apart from other forms of targeting?

Top filtery goodness of the last few days:

• The Supreme Court is looking into data mining and seems to be leaning toward protecting pharmaceutical companies’ right to do it: http://www.washingtonpost.com/politics/supreme-court-skeptical-that-prescription-data-laws-dont-violate-free-speech/2011/04/26/AFDKIzsE_story.html
• Turns out that whole iPhone tracking scandal was a bug? http://tech.fortune.cnn.com/2011/04/27/locationgate-was-a-bug-says-apple/
• Microsoft is working to create web browser standards to protect user privacy. Hopefully money won’t get in the way of good tech: http://www.technologyreview.com/computing/37442/?p1=A1&a=f
• CNN blogger looks into the many exciting possibilities of news aggregation, but offers a cautious conclusion: http://tech.fortune.cnn.com/2011/04/26/news-curation-finally-social-medias-killer-app/?section=magazines_fortune

Eli’s book, of course, is about what happens when personalization algorithms filter the information we see – but he’s also frank about the fact that we do a lot of information filtering on our own, without the aid of online bots. For the most part, we don’t go to Drudge or HuffPo because an algorithm leads us there via our friend’s Facebook feed; we actively choose our information sources based on whether they share our worldview.

But the self-reinforcing forces don’t stop there. Even if we were to skim an article that disagrees with our beliefs, we’re unlikely to read it with an open mind. Instead we’ll cherry pick the information that supports our position, and pick apart everything else. (In one famous study, subjects who read identical papers supporting and contesting the death penalty tended to strengthen their previous views, regardless of whether they were for or against capital punishment.) And if the argument comes from a source we distrust – well, the odds are even greater that the only facts that will filter through are those that jive with our viewpoint.

Youtube seems to know that, likewise, a Republican politician has a long shot of winning any converts from the Democratic party when it comes to a debate on the issues (and vice versa). In a new civic-minded venture with Congress, the Google subsidiary will host pro-con policy face-offs between US senators starting May 2. To get past our partisan filters and increase the chances we’ll give both sides a fair hearing, the senators won’t be identified by by party.

The project planners are banking on viewers not being able to tell a Jim DeMint from a Dick Durbin (wisely so). But although it can remove the “messenger” bias, there’s not much Youtube can do to stop us from sifting out disconfirming information once we know the upshot of the “message.” Nonetheless, getting past two filters – by showing us two sides of an issue and (partly) cloaking the source – is a promising step in the right direction.