Improving Northwind is a good idea. However, let's remember that databases are used in a lot of diverse ways. So, let's not make the mistake of thinking that there can be one database that best demonstrates all of the recent developments in db technology and theory. For example, we may need a database for "Voyager Transportation" to show the power of geospatial data.

Having said that, I recall a number of debates on this forum with topics like "why bother with CLR?" (stated in more polite terms) and "why bother with XML in a database" (I think the editorial writer for that one was a little more blatent in his dislike for XML). If Microsoft is going to put the resources into adding features like these and more to SQL Server, perhaps it would behoove them to come up with some examples of how these technologies can shine in a proper setting?

If it's possible to put multiple developments into one database, then great. Let's just be careful as Database "experts" to avoid the mistakes that we always criticize others for doing, such as creating a swiss army knife database that contains a lot of technology, but which doesn't really make any sense in the real world.