Tracking Users Via Latency > waaaait a second

waaaait a second

network-latency-discovery.l.google.com

It would seem as though you would have to have an idea how many data points they were working with, how finely tuned they are, and how they system worked.

Lets assume that your average monkey with a computer and access to traceroute.org could figure out what city someone lived in.

Then assume (for the point im trying to make, that this is feasible) that these other people have 1000 datapoints in this city. You could call that crazy but say if someone really well integrated into society such as AOL, Kazaa, Comcast etc is involved its very feasible that its many times that.

Then take your average ping response of 1ms, 2ms etc. and make it 1.0334433, 2.0023444 etc. so they actually look like something that could be considered a valuable piece of data.

Then assume every datapoint is constantly hitting other network points that could be either known distances away or not known distances away to know what your average network looks like on 1000 different average days.

If the measuring is accurate enough (im pretty sure its not running microsoft's ping tool), and enough datapoints are involved from known locations (outsourced in major cities to cable providers via integration into cable modems etc.), I'd have to assume that they'd be able to get a pretty good handle on where computer X resides at.

My little theory obviously has some holes but its not exactly cold fusion were talking about here.

-Brendan

reply to china crisis
It's part of Google's 'desktop', that thingy what tells you the weather in your zip code.

So, if enough people on enough networks are running GDesk and getting the weather, who knows what those bright folks in marketing can do with all that data.