*sigh* Ok, let me try to explain this more carefully for you.
A user with an iPod Touch (a device that has a WiFi radio that can detect WiFi base stations, and also use one for internet access, but no GPS chip at all) is joined to an open network at a convention. He opens Google Maps, pulling in data tiles for the area nearby, and these get cached to the devices flash memory. Also cached to the device flash memory are the latitude and longitude of all the base stations in the area, say for a 2 mile radius (just as an example).
The user then walks away from the convention center, and the iPod drops off the open WiFi network, thus the device has no active data connection. It does however have the cached map tiles, and information about other WiFi networks. Using triangulation, the iPod can still place a blue location dot on a map drawn from the cached tiles saved earlier. It's not as accurate as GPS, but possibly good enough, especially if on foot. The more WiFi access points it can see, even if they are secured, gives the device more reference points to calculate location.
Along the way, the user takes a photo of something interesting. Due to the cached WiFi data, the photo ends up having pretty accurate latitude and longitude information added to the EXIF data, assuming the user enables geotagging of their photos.
Thats the part about why this cache is useful for the iPod Touch and iPads with WiFi only, with no 3G or GPS chips inside.
Now, the second part, why not use the device saying "I'm here". I'm assuming you mean the GPS chip. A pure GPS system takes a while to lock onto a users location from a cold start, and even a warm start can take a while. GPS chips also consume a decent bit of power, especially during cold start situations and while trying to obtain a positive position lock. One way to accelerate a lock is to already have a rough idea of where the device is. This speeds up the lock as it can then figure out what the GPS constellation looks like currently, and pick up the right signals to then get a better position. I'm not going to go too in depth on how GPS works, as Wikipedia can fill you in there.
The basic point is that GPS takes a lot of power, WiFi scanning takes less power, and cell tower scanning is even less power (due to the phone already having to keep track of towers to make sure the phone part works). Using cell tower triangulation results in a pretty poor location result, due to the low density of towers. WiFi positioning can be far more accurate, due to the smaller radius of the signal, and the higher density in most cities. And GPS gives the best location since thats what it was designed to do. By balancing all three and using cached data, you can provide a user useful location information quickly, without causing too much power draw or data consumption (counting against monthly quotas, and adding latency to the location calculation).