This isn't an isolated problem; it comes up in other contexts as well. Consider the default zero-zero GPS coordinates (and use an app for plane or shipping tracking to look at it). It's about 200 miles off the coast of Africa and there are thousands of dodgy transponders identifying various vessels as being located there.
And then there was this deplorable incident ( http://www.freerepublic.com/focus/news/653846/posts ) in which changing the batteries in a special forces' air controller's GPS receiver caused it to reset and display its current location rather than the previously dialled in coordinates of a target, resulting in a fatal friendly fire incident.
Moral of story; if you're building a geolocation database or application of any kind it's a really good idea to ensure that the default values it returns in event of a reset or a query with inadequate input parameters points somewhere harmless or pointedly asks, "are you sure about that?"
> Consider the default zero-zero GPS coordinates (and use an app for plane or shipping tracking to look at it). It's about 200 miles off the coast of Africa and there are thousands of dodgy transponders identifying various vessels as being located there.
There actually is a weather bouy at 0, 0. It's mapped in OpenStreetMap with a note of "This is actually here, it's no a bug"
I think the distinction between nullable types and the non-nullable types you speak of is fairly narrow, you've just made it type safe (which isn't a bad thing!).
I don't necessarily disagree, what I don't understand is the parent comment's use of the word "need" in "Both need to be permitted". It makes me think there is something nullable types can do that non-nullable types can't do, and I'm just not sure what that is. Something like "it is fine to permit both" would make sense to me.
I guess maybe it's because in a language with no value types (like anything on the jvm), using Optionals everywhere requires a lot more objects to be allocated, so in allocation-sensitive code it might make more sense to use nulls instead.
I still ask the question, why use default GPS coordinates? Even after moving the default to a nearby lake, there's still a large chance that police and investigators are going to prowl around the lake trying to find whoever's house lays claim to the center of the lake. Why not have defaults return invalid or perhaps a square box instead of a point? Moving the point around isn't going to fix the root of the problem.
I think the problem is not that, but that the people using geolocating ips do not understand that there is "accuracy" metadata associated with it. Or maybe maxmind (or whoever is using maxmind) are ignoring the accuracy for that specific location. To geolocate someone with just an ip, you should never expect it to be more than "country level" accurate.
I can def see detective Jonhson trying to find child pornographer X with a warrant to get their location based on an ip. The company doing the geolocation might or might not tell them: this is the gps coords, but with an error of a few thousand kms.
Now, if the Arnold's ever want to do something "bad", their situation could've been the perfect cover up :-)
> you should never expect it to be more than "country level" accurate.
Well if we're talking about safe expectations... there are no expectations. Pretty sure it'd not even get the planet right for browsers from the international space station, or the continent if your IP is newly registered (e.g. IPv6, but even for IPv4 this still happens).
Cities, in my experience, are often correct and are usually the closest you get.
I think the problem isn't that it's "invalid" but that the number returned should be an area and not a point - they have enough information to say the IP is located in the US. It's somewhat reasonable to return this as a center + a radius of certainty, saying, "It's somewhere in this circle" - I'm not sure if Maxmind returns a radius, but I wouldn't put it past the police to see a point in the center of a giant circle and say, "OK, let's go to that point" - not realizing the point of the circle.
In reality, it should probably return a shape file of some sort.
MaxMind does not say their IP info is good to even a city BLOCK level, much less an address. Law enforcement is using the data all wrong.
I've used max mind 5 days a week for years and have never seen this effect.
People watch too much CSI. That's not Maxminds fault.
LEO should be contacting the ISP to get warrants for the customer/ip records. The ISP should know who had what IP when. And the ISP knows where that customer lives because of reasons.
tl;dr - maxmind did nothing wrong, LEO watches too much CSI
What do you mean maxmind did nothing wrong? If their data isn't accurate to an address, they shouldn't be returning an address, especially to law enforcement
TLDR: MaxMind uses the coordinates of their home as the default view. Meaning when law enforcement enters an address but fails to actually submit it, it shows their home.
Yeah whoever is coming there has some false ideas about how accurate IP addresses map to physical addresses based on publicly available data. If anyone could find my home address based on my IP address, you bet I'd be using VPN 24/7.
Maxmind could have moved it to a lake or something, but other than that I see no fault on their part.
http://fusion.net/story/287592/internet-mapping-glitch-kansa...
https://news.ycombinator.com/item?id=11466849