It is easier to implement Mastercoin on the blockchain. The entire concept of mining "vanishes", or actually is satisfied by Bitcoin miners. This means the security of Mastercoin is on par with Bitcoin (up to potential bugs) ... there is no way to 51% attack it.
The same could be implemented using merged mining, but no matter how you look at it, merged mining is technically a more complicated solution, because it requires an entirely new blockchain one needs to reason about. You need to think about the hashrate of this separate blockchain and to encourage miners to merge-mine it, things that are just not needed when you directly encode things into the Bitcoin blockchain. As far as software development goes, I'm super lazy - I do not like to do something complicated if I can do something simple that serves the same purpose.
So your entire reasoning is that, while merged mining is in fact a solution, you are too lazy to do it? You need a specialized client already, so why not mastercoin-qt. Sheesh. It takes almost zero effort to start an altchain, evidenced by the rash of them already.
Your argument is weak. Let's face it, Willet just wants to do it this way, there is no other reasoning other than, "I want to, I can, so I will".