Confusing an estimate with reality

This post at kottke.org illustrates a problem with the user interface on many mapping applications. The application is supposed to show you where your phone is, but when it can’t determine the position accurately, it attempts to find a general location.

In the case of Sprint users in North Las Vegas, the “general location” is  centred on the home of Wayne Dobson. Instead of showing a diffuse area, the software is somehow displaying a point, with a street address, and that’s Wayne’s house.

This causes problems for Wayne when people show up looking for their phones. It also caused him trouble when the police show up — because someone called them from a cell phone, and the location given is his address.

This strikes me as very odd indeed; not only average users looking for their lost mobile phones, but professional dispatchers are being fooled by some sort of position reporting that doesn’t indicate the error in the position. And this has been going on since 2011!

Can you imagine the waste of resources when the police go to the wrong address? Not to mention the fact that they are not responding to the right address where someone was legitimately in distress.

The really strange thing is the media reporting on this seem to take it in stride as an artefact of a “am imperfect system” — somehow this is just a position reporting glitch.

The fact is, all GPS systems include error and should report how accurate they are so users can understand that the “point” on a map is not really a point, it’s an area of probability. When that area is really large, it means the position has lots of area and should be taken, as they say, with a grain of salt.

I think people are being fooled by user interfaces that seem to indicate certainty. Small changes in a UI could make a big difference on how people understand the technology. Unfortunately, GPS manufacturers sell certainty, and a UI that is less than certain isn’t a feature that sells units.

Posted in Blog Entry Tagged with: , , , ,
2 comments on “Confusing an estimate with reality
  1. Maybe you have some insight on what’s happening, but I assume from the fact that this is just with Sprint phones, and all to a specific location, that what’s happening is normal GPS services are not getting a lock, and they send an aGPS request to the cellular network, and for some reason the network is routinely reporting a single location?

    What do you suspect as the failure mode causing what I would expect to be a random scattering of best-guess locations to all end up being one location?

    • I think what you describe is exactly what’s happening.

      From using some positioning API’s and even some GPS units, they regularly report position data without the accompanying error estimates. Even the GPS units we use for SAR are consumer grade, and while they report the error on screen they do not record it. When we end up displaying it on a map, it’s context free and very hard to tell what the actual conditions were.

      Going further, the position reporting system may or may not send the error, but the dispatchers and public are almost certainly several layers of abstraction/api calles away from the actual system. Their displays, which may or may not show the error, are also probably showing a “point” in the center much like how Google Maps on iPhone and Android shows a point in the center of a blue circle. The blue circle, and it;s size, is the thing to pay attention to. The point is completely misleading.

      In summary, I think either the error information is not being displayed, or is being displayed in a way that is too easy to misinterpret.

Leave a Reply

BlueToque Software
BlueToque Software
Software Design, Architecture and Consulting

TrueNorth Geospatial
TrueNorth Geospatial
Mapping Software for Backcountry Professionals

YourLo.ca/tion
YourLo.ca/tion
Locate people using their SmartPhones

Older stuff

Twitter