Home » CurrentHeader, Digital Insight

Speculation: Could Google’s ‘Payload Data’ Be A Broadband Map?

By 28 May 2010 3 Comments

Please note: This post is purely conjecture.  I have no evidence to suggest the theory suggested is true.  This post is offered purely as speculation and should not be relied upon as factual for any reporting.

I’ve been discussing Google’s improperly collected payload data, and their reluctance to hand it over, with a lot of people.  Most of the discussion comes down to the legal implications of violating privacy laws twice – once by gathering information surreptitiously, and then by handing that ill-gotten data over to a third party.  But occasionally that discussion comes down to a question of what is in the data.

When countries have asked Google to destroy the data, they have been more than happy to comply.  When asked to produce the data, Google has hedged.  Is that simply a function of legal compliance? Or could there be some “there” there?

To find an answer, I asked the following questions:

  • How much data could Google actually gather from your network while driving past the house at 20-30 miles per hour?
  • What kind of data would actually be useful?

The answers to the first question is “not much.”  That actually makes the answer to the second question more interesting.  Google has described the data collected as “only fragments of payload data.”  So what sort of fragmented data would be useful?

I have one possible theory, but no evidence to specifically suggest this is true.  It’s just the only explanation that I can come up with for something useful in a very small amount of data.

As the street view car passes by, the street view car could detect an open connection.  It could then attempt to connect.  Assuming a successful connection, it could move a small piece of data upstream/downstream over that connection.  It would only have a second or two at best, but could move a small piece of data in that time.

What would that give Google? Well, first, it could tell them who your ISP is.  Second, it could give them a very rough sense of your connection speed. As explained to me by an engineering friend, the accuracy of that speed test would be “worst-case”.  There is just not enough time to do a thorough test with the car moving at speed.

However, when aggregated with many readings in the same geographic area (especially many readings from what’s likely the same ISP or ISPs), a fuller picture of coverage, and coverage area could be obtained.

Google could be using its street view cars to map out broadband service areas and capabilities.  They’re driving through the most heavily populated areas of the world.  They’re scanning for networks and grabbing traffic from them.  An analysis of that traffic would give you a pretty good sense of ISP footprints and speeds.

Pair that with a project to do something really ambitious (like a fiber rollout, perhaps), and you would have a competitive advantage in terms of where to build, where – precisely – the best customers would be, what they’re currently getting from their ISP.

Google has said they have no interest in competing with ISPs, and the fiber initiative is simply an experiment.  However, a robust picture of ISP plant and capacity could give you the upperhand in competition.

Gathering that map by spying on open wi-fi connections could prove to be a PR nightmare, however.


  • Wes Felter said:

    They were passively sniffing, not associating.

  • larry seltzer said:

    Your theory really just isn’t plausible, assuming you believe all of Google’s descriptions of their procedures. If in fact the car is on the move and it changes channels several times a second they wouldn’t often have enough time to connect and transmit data.

    It’s far more believable that, as the other commenter says, they were just passively recording network traffic.

    To me this says more about how difficult it is to comply with privacy regulations and how people go nuts every time there’s an innocuous problem like this. Really, there’s every reason to believe that this was accidental and that nobody was harmed.

  • Michael Turk (author) said:

    As I said, I have no evidence that any of this happened. It is just a possible use of their technology given the time available.

    Even your comment is based on the condition that “you believe all of Google’s descriptions of their procedures.”

    The car has to travel during the bright of day given the shutter speed necessary to grab crystal clear images at 30mph. Also, we’re more likely talking about residential connections, because businesses are more likely secured than open. They also said the user had to be actively using the connection.

    Yet, they gathered 600GB of data from people at home, during the day, likely on weekdays, actively using the network, and in miliseconds of exposure?

    That seems like a staggering amount of data given the relative difficulty of meeting all of those conditions.