Theorem 1 - It is impossible for a message to be both extremely clean and extremely dirty at the same time
According to the Spam Curve, the bulk concentration of messages are the ones that contain a lot of spammy content (extremely dirty) or they contain plenty of legitimate content (extremely clean). We all know what dirty messages look like, they contain obfus.cated words, $#%& curses all over the place, and OTHER OBVIOUSLY spammy content. Lots of messages are like this and are easier to filter.
Legitimate messages are those which are easily recognized as legitimate. They often contain content like name and address information, phrases that rarely, if ever, occur in spam messages or links that spammers never use. We can instantly recognize these types of messages as not at all spammy and a decent spam filter ought to easily be able to tell the difference.
Because of this property, a spam message can never be both extremely clean and extremely dirty at the same time. Simply by looking at the chart we can see that this is a mathematical impossibility. Now, a naysayer may come along and try to contradict this by saying that there are exceptions to this rule. A very legitimate message can contain some extremely dirty content inserted somewhere in the message, either deliberately or accidentally (one example is forwarded spam, another is a non-related reply to a dirty joke). This, the naysayer might claim, contradicts the very first theorem because the message is now both very clean (it has the non-related and possibly business-related) content but also the very dirty part, perhaps the spam attachment or a dirty joke. See? Very dirty and very clean at the same time.
However, this contradicts the inherent nature of a very clean message - that it contains no dirty content. Recall from above that clean messages are easily recognized as legitimate. A legitimate message that contains dirty content somewhere in the message is no longer a clean message, it is a message that contains both and now falls along the overlap portion of the Spam Curve. If you have a solution with a pH of 1 and another solution with a pH of 13 and mix them together, you don't have a solution with a pH of 1 and 13 at the same time, you have a diluted solution with a pH of around 7. The same is true of email, a pure message that is diluted with dirty content is no longer pure. It is now diluted. This demonstrates theorem 1 - It is impossible for a message to be both extremely clean and extremely dirty at the same time.
By the way, I am aware that the concentration of the solutions in my example would need to be equal, etc, etc, but you all understand my point.