Actually, I find it interesting to note that the Bayesian spam filter I use catches these types of emails All Day Long (tm). It seemed a curious thing to me, wondering how it was picking these out from more legitimate email. I started analyzing the emails and realized the highest spam words were being grabbed from the headers. Sure, headers can easily be modified, but most spammers apparently aren't that sophisticated. They use common tools with standard headers (usually advertising the tool they are using), which are very easy for the filter to catch.
As the spammers realize this and start using tools that are harder to catch, there will still be things like MTA versions and hostnames added to the emails along the path. Perhaps certain MTAs with bad default options will begin to stand out as likely spam targets. Perhaps certain IP blocks will begin to stand out, also. Who knows? The great thing is I won't have to think about this. The filter will figure it out automatically.