It's been too hot and rainy here in Thailand for extensive bicycle mapping, so I've been trying to clean up TIGER data from the US. It seems to be a mess in many areas (and just great where users are active).
I've spent most of a month or so aligning OSM roads in small to medium sized cities in the US. My criteria is just that the city has few or no other users or GPX traces, and is a mess. I just align them to Bing, and the results by comparing in Transparent Map are nearly perfect (The errors seem to be mine, and Google Map has similar inconsistencies).
Because of the random and careless placement of nodes, many of these areas had not a single navigable road depicted on OSM. Often I have to count from a landmark to correctly place them. First, I thought this must be vandalism, but no, it was TIGER. The topology seems fairly reliable tho, once you find a starting point.
This is an embarrassment to OSM, but not so hard to fix (Just tedious). There is some worry on blogs about the quality of the aerial images available to OSM. My experience suggests that what ever the image quality (and I've found nothing wrong), it is dramatically better than what is on the map now.
I think this is an important and time sensitive issue. One GPS manufacturer offers an OSM based unit. If they sell any in much of the US, the customers will never use OSM again. Portland, Oregon has adopted OSM for their transit planner website. This is a major city risking a major plan on the Map. If it works, other cash-strapped cities may follow.
Summary: The Map can be hugely improved without waiting for individual users to bike or drive every road. Most of the cities I've worked on have had no traces or edits in the history of the project. In short, use the largely correct TIGER data for street names and topology, and Bing for alignment.
Wish List: A way to offer tiles to volunteers to clean up, so the work could progress in an organized manner.
While I've done a bunch of edits, they are mostly the same things over and over. I don't know the intricacies of OSM and am probably missing much. Any constructive criticism appreciated.