Another question that puzzles me is the one below:

Your team decides it needs to increase the skills of its individual members beyond their primary skill set. Which tactic is likely to be the least effective?
A. Sending members on training courses
B. Pair developing (e.g., pair programming)
C. On-the-job mentoring from other team members
D. Random allocation of tasks to team members

Answer sheet says it's B, but if you're pairing-up with another developer wouldn't you learn from each other? Also, why would D not be right because if you give random allocation of tasks to the team members and they get something they don't know how would that be effective? Wouldn't this be the least effective?

Thanks.