[ih] How an internet mapping glitch turned a random Kansas farm into a digital hell (Fusion)

[Bill Ricker]

On Thu, Oct 2, 2025 at 3:59 PM John Levine via Internet-history <
internet-history at elists.isoc.org> wrote:

> It appears that Eberhard W Lisse via Internet-history <el at lisse.na> said:
> >While not uninteresting, where is the IH relevance?
>
> Their problem was that one of the large IP address geolocation
> services had a default value for unknown networks that happened to be
> their house, and a lot of people were sure that their lost or stolen
> whatever they'd located by IP lookup was there.
>

Not exactly "Default."

The normal blank value returned for NOT FOUND when geocoding without
separate success/fail flag [as good protocol design would have!] is
(latitude, longitude)=(0.0,0.0), which is the home of the buoy at 0° N, 0°
E-W in the Gulf of Guinea.

(Yes, there's now a Buoy there.
So it's now *_sometimes_* a valid value, but rarely.
(0,0), was nicknamed "Null Island" because of the geocoding default issue
long before the buoy, which is officially "Soul buoy", but gets called Null
Island still too.
(-99.99, -99.99) might have been a better albeit COBOLish NULL value which
would be out-of-range to use for mapping or directions and one hopes error
eventually in a vaguely useful way. Now they tell me!)

The problem for this farm is if  geocoding the IP address resolved to *USA*
but not *_otherwise*_ differentiated, such as if it were a dynamic address
for a national provider/reseller, the geocoding does the same as it does
for any much smaller region name: it returned the centroid of the *named
region*, which by their calculation of the centroid of the USA was on this
very farm.
(Other centroid calculations get different centroids for USA or Lower 48;
the old size-of-ruler-problem.)

Alas it returned excess digits of precision without a (± 1500 miles,
±1000miles)  error range to show the lack of precision, and since it pinned
a specific domicile, was (mis)understood to be a definite
IP-address-to-house-address match.

(While we know people with single family homes with static IPv4 addresses,
that's the exception, and someone should've told the cops or at least the
D.A.s & Judges before they started geocoding IP addresses for search
warrants. _We_ don't believe computers are infallible because we have to
fix them, but so many others are gullible and accept GIGO answers at face
value.)


I believe the fix was to move the default to the middle of a nearby lake.  I
> expected some enterprising kid would set up a kiosk renting rowboats and
> snorkels.


Sounds vaguely familiar.

One hopes they arranged for _all_ centroids to land in parks or ponds and
not homes, not just the USA centroid that complained.

(The genealogy application that I use uses 42.3 N, 71.8 W as centroid of
Massachusetts; that is moderately safely in the rear parking of a Worcester
Post Office (01606). Wikipedia uses 42.3 N, 72.0 W as Massachusetts; that
is in the woods by the pond on the grounds of St Joesph's Abbey, home of
Trappist Preserves, but far from both the buildings and the nearest state
road. So rather obviously a centroid and not a definite geolocated 'hit'.
Another algorithm could pick centroid anywhere between or somewhat beyond
those points.)

-- 
Bill Ricker
bill.n1vux at gmail.com
https://www.linkedin.com/in/n1vux