Critical chain cuts down on the extensive padding put on most tasks in regular project management. Critical chain starts at the end of

click to enlarge
the project and works backward. Based on previous experiences, estimates can be made on each task to see how long it will take. Then, some space is added between projects to give some leeway on the estimates.
These estimates are based on best guess estimates and the probability of that task being done within that time frame. Then, the available resources are assigned to each task.
Critical chain focuses on cutting the buffers (time between tasks) down to 50 percent. This puts immediacy on the task at hand. It also cuts down on multitasking. If a task needs to be done before another task can be started, team members cannot jump between tasks.
Multitasking can be very detrimental to the overall project. Team members are stretched thin trying to work on three or four projects at once, basically ensuring that the overall project will not be done on time.
In most other methods, people are not rewarded for getting their task done early since other tasks needed for the critical path are still behind schedule. Critical chain eliminates this problem.
Since the buffers are cut by 50 percent, the method assumes that some projects will get done early while others will take longer. But, when one task is completed early, the next task can be started. This means the tasks that get done early and those that get done late, will average out and the project will be completed on time.