Dec 8, 2012 · 4 minutes

Under Larry Page, Google has developed a dive-first-learn-to-swim-later strategy, a willingness to embrace more risk than most companies making $40 billion a year in revenue would sensibly stomach. There was the push to design Google's empire of features around Google+, when most of its ad dollars were still tied to search. Then there was the $12.5 billion purchase of Motorola Mobiliy, when it wasn't clear why Google needed to manufacture phones. (It still isn't.)

But what may prove to be Google's biggest -- and potentially dumbest, or potentially most brilliant -- roll of the dice is Google Fiber. In February 2010, Google put out a request for proposals, asking US cities to host the company's quirky experiments in building one-gigabit-per-second fiber connections to homes.

Although Google received inquiries from 1,100 communities, many people didn't realize how serious Google was about executing its crazy plan until this summer, when it began encouraging residents and companies to start signing up for Internet access fast enough to download a hi-def movie in seven seconds. With a free Nexus 7 for a remote. And a free terabyte of data storage. And all of this for only $70 a month.

At first, Google Fiber inspired more questions than it answered. Would enough people sign up for it? Would they really want all that speed if they're just posting on Facebook? Would there be compelling video content, and an interface to manage it? Could Google avoid glitches and create a smooth customer support system? And most of all, could Google afford to do this in more than a few cities?

A month ago, Google began installing fiber into Kansas City homes. Since then, some of these questions have begun to get some clearer answers. People began declaring their rapture on Facebook and posting screen shots of speed-test results to Twitter. It turned out the 1 Gbps speed promised by Google wasn't exactly true. Some people were getting only 700 Mbps. Which, you know, is only 70 times the typical broadband connection in the US.

While the interface and video-offerings issues still need to be ironed out, the rollout of Google Fiber has gone smoothly. The experiment, as experiments go, is so far a success. Google is ready to roll out Fiber to new cities, and it's actively hiring for dozens of jobs to keep the Fiber train moving.

One analyst, Rich Greenfield at BTIG Research, argued in a recent note that Google Fiber will not only accelerate rapidly, it will change consumer habits. This raises new questions that are potentially more interesting in the long term. Just as the leap from dial-up to broadband connections allowed new behaviors and business models on the Web, the onset of connections of even a hundred or so megabits per second could be just as dramatic. That would create a field day for developers building apps much more powerful than they are today.

Of course, all of this is still just a possiblity, and far from a reality for most Americans. And sure enough, there are also people coming out to pour cold water onto the prospect for a new era of super-broadband. At an industry conference yesterday, Time Warner Cable's COO Rob Marcus brushed off Google as a threat in Kansas City, saying the company has seen no need to increase its Internet access speeds. But that it could if it had to.

Marcus seems blithely ignorant of widespread consumer dissatisfaction. Rory Maher, an analyst at Capstone Investments, scanned 5,000 comments on Facebook about Google Fiber. What he found was not so much an enthusiasm for Google per se as a seething resentment toward the broadband incumbents. Pricing was an especially sore point. A few were considering moving to Kansas City. One success of Google's experiment is that it proved what broadband subscribers have long suspected: Faster access can in fact be delivered at much less cost.

Another glass of cold water came from the research desks at Goldman Sachs, which issued a report today estimating that rolling Google Fiber out to the entire US would require at least $140 billion in capital spending. Google only has only $46 billion in cash and taking out loans to cover the balance would crush it with debt. So nobody's really expecting Google to handle the buildout on its own.

But this takes us back to the reason Google set up the Fiber experiment to begin with. In June 2009, the company submitted comments in support of the FCC's National Broadband Plan, which would later recommend bringing 100+ Mbps Internet speeds to at least 100 million U.S. homes (there were 115 million households in 2010) by 2020.

As with many federal initiatives, there are flaws with the National Broadband Plan. But the plan is intended to support Internet businesses building the Internet. Google Fiber was never meant to be only the company's entry into the ISP business. It was meant to show how wonky policy can become a coveted consumer service. What Google built in Kansas City is a working model for the future Internet. And that's the biggest reason why it's worth our attention.