The problem with basing your exposure on an average is that if the contrast range in the scene exceeds "normal," then the middle value you end up basing your exposure on is pretty high. This results in underexposing the low values and the corresponding loss of detail. In other words, you are no longer "exposing for the shadows," which is a better method than averaging.
Right on queue Doremus, you were correct.

I did a test exposure yesterday using the average of 3 values as I have never tried this before.

I metered the deepest part of the trees, the middle value on the bridge and the brightest part of the sky and then hit the AV button and used that for the exposure.
You were correct, the image is under-exposed and the shadow areas have indeed lost detail.

Click image for larger version.  Name: Sprot-Bridge-test-Average.jpg  Views: 53  Size: 56.5 KB  ID: 179287

Apologies for allowing the cable release to enter the frame to busy trying to mentally see what the average would give me