🚀 go-pugleaf

>Where I live over 50% oh households turn off their routers when they are 
>not at home and at present I am staying at relatives, about 200 miles 
>north of where I live, and that figure has increased to about 80%. I 
>would also say that with people changing their ISP options and 
>consequently their SSID's, most people never change their ISP's unique 
>SSID's but everytime the router or ISP is changed the SSID changes, 
>that their has been a 40% change in the last 18 months and I doubt 
>Google are going to renew their database that frequently so I am not 
>convinced about using consumer wifi as a location device in the UK. It 
>would probably work with wifi hotspots which tend not to change and are 
>near to your "fish and chip shops" but I see no reason for them to be 
>recording wifi networks on housing estates which are a couple of miles 
>away from anything.

You'd have to think like a programmer to realize these things, but:

1.  MAC addresses (Bob Metcalfe et al & Std Bodies) are used, not SSIDs.
2.  When you are a user of the data, you also send the data back to the database
    about what you find.  So, you find out MAC address A and B are in
    the known database, and MAC address C D E F G and H aren't around
    that are in the database, and MAC address I J K L are around but
    not in the database.  You send an update to the database of what
    you found.  The database and you do probabilistic calculations to
    figure out if MAC addresses A and B simultaneously moved or if
    you're actually where the database has them marked.  This has to
    do with tracking the location the set of all MAC addresses across
    time and place.
3.  It's not just the wifi routers:  it's the clients, too, so that doubles
    the number of stations to store info about.

So, the 50% to 80% of households turning off their Wifi routers or
changing their SSIDs doesn't really stop the database from working.
Coupling it with real-world use (constant update, GPS backup data,
people who don't turn off their wifi, etc.), the database really is
very useful for what it does.  I can see that in some pockets it isn't
as useful, but for the most part, it is very useful.

Google driving about is basically to seed the database faster than
GPS-enabled responders can, but with GPS response updating the
database, it's basically not necessary for the Google seed.  It's just
a quick jump-start to the whole process.

Of course, if it's programmed wrong, then it doesn't have to work
right or well.