[wisty]

A basic map application just requires data, projections (which is a solved problem, if a little tedious), and a rendering system. Path finding is also basically solved (it's Algorithms 101, though you need something a bit more sophisticated to make it scale). A basic mapping app can be hacked up in a week. A good mapping app is obviously much harder, but it's still doable.

But like taligent said, it's data that's the real problem. Mapping data tends to be dirty and heterogeneous. Do you have a point, or a polygon? You're in trouble if you just have a street address (geocoding can be very hit and miss). Metadata (like the projection) can be missing. Locations can be slightly wrong, and you'll draw a highway running through a shopping mall, or connect streets which don't quite connect, or have gaps in a street because it changes street names and there's a tiny gap between the two streets (which doesn't exist in the real world). How do you normalise the field names? How do you even get the data? Once you've got everything into your database, you move onto the next city / state / country.

[zaphar]

Lets see if I understand you right. Mapping is not hard. You just have to get good data. Getting good data is hard though. And getting good data is part of doing Mapping. So basically you said Mapping wasn't hard and then said that a crucial part of Mapping was really hard.

All of which is a roundabout way of contradicting? yourself.

[taligent]

This answers my point better. The fact is that many of these mapping companies that even Google still licenses to this day have been driving around countries in some cases for decades.

That is the real hard part. Trying to obtain all that data. Because every mistake is potentially one person complaining loudly on the internet.