I'm reading the paper “Collision-Resistant Hashing? Towards Making UOWHFs Practical” which states:
While it might be easy to find a collision $M,M'$ in $F_K$ by making both $M,M'$ depend on $K$ (Any Collision Resistance), the adversary may be unable to find collisions if she is forced to commit to one point of the collision before seeing $K$. We call this weakened notion of security target collision resistance TCR
nb: $K$ is a key used to a hash function $F$.
My question is: Why then TCR is a weakened notion if the article says that it is easier to find a collision $M,M'$ than TCR?