Yes, I agree. As a mathematician, I spend lots of time trying to find the correct definitions with only a rough idea what I'm trying to prove (the theorem).
Conversely, in theoretical cs, there are many longstanding open problems. People who make progress on these problems are taken seriously and listened to precisely because they have new techniques and novel approaches that have not been considered previously - i.e. the theorems & definitions are well understood, but the proofs are the significant element.