Think of it this way:

Visual Studio 2005 ("Whidbey") is intended to provide the best development
environment for the .NET Framework 2.0.

Visual Studio "Orcas" builds upon Visual Studio 2005 to provide improvements
for developers building applications for Longhorn. This is due to the fact
that the .NET Framework 2.0 is basically the core programmability engine at
the core of WinFX, the primary API set for Longhorn.

The reason Microsoft has posted this roadmap material is in response to
customer requests for more transparency. For example, we will launch Visual
Studio 2005 in early 2005. However, since Longhorn is due to launch less
than 2 years later, there is concern over whether Microsoft would launch a
new Visual Studio and obsolete the 2005 release. By providing this
transparency, the goal at hand is to be clear to developers that 2005 is the
next version and the foundation for the version to follow.

By comparison, other tool vendors don't generally plan this far out. As a
result, the typical enterprise that makes a developer tools/platform
decision every 3-5 years doesn't have a good idea about where these tools
will be during that time.

I wouldn't put much stock in the rumors regarding CASE and fully graphical
development. Although there are a lot of places where boilerplate coding can
be cut out to save time, the fact is that tools won't be able to completely
eliminate manual coding anytime soon. If you take a look at products like
BizTalk you can see where manual integration coding has been eliminated
(partially thanks to standards like Web services and  practices like SOA).
Other places where coding has been reduced revolves around UI and typical
boilerplate stuff, and the 70% code reduction ASP.NET 2.0 introduces over
ASP.NET 1.1 is an example of how that gets better. However, sophisticated
applications (which all are at some level or another) will still require
hands-on effort from developers for the forseeable future.