Susan Smith Nash raises some very good points about learning objects - what are they? why would anyone care? how would someone reuse them?
CAREO is mentioned specifically, and the concerns with it are entirely valid. It should be noted that CAREO and its ilk are from the first generation of learning object repositories - a necessary step in the evolution of the concept - and we've got some stuff that begins to address many of these concerns (Pachyderm, APOLLO, etc...).
This "second generation" learning object repository stuff is just starting to see the light of day - and even that isn't anywhere near where we want to end up. It's all just baby steps along the way...
Disclosure: I built a good chunk of CAREO, and am pretty deeply involved with the development of APOLLO and Pachyderm.