Iteration planning: hours of tasks or number of tasks?

I recently had a discussion with another Scrum Master. He participated as a chicken in my sprint planning meeting.

When the team had finished the breakdown of user stories into tasks, I wanted to get estimates in working hours for every single tasks of the sprint. The other Scrum Master wondered why I insisted on getting those hours. He would have just worked with the number of tasks in the sprint rather than put more effort in getting those estimates. In his eyes that would be sufficient to track progress and draw an iteration burndown.

In my point of view it is not enough to have the number of tasks solely due to following reasons:

• task sizes are unknown and not comparable
• it is difficult for the team to choose open tasks for the next day without knowing an estimated effort in hours
• it is impossible for the team to honestly commit to a story on the basis of story points (or ideal days) only
• the iteration burndown will get more realistic and shows "real work done"

Finally an iteration contains a more detailed scope than a release so more detailed estimating and planning has to be applied.
Bibliographic recommendation on this topic: Mike Cohn - Agile Estimating and Planning (http://www.amazon.com/Agile-Estimating-Planning-Mike-Cohn/dp/0131479415)