Question

According to Frederick Brooks (The Mythical Man-Month), how should the distribution of effort for each of the primary phases (as fractions of the total) be scheduled in a software development project? How does this differ from conventional scheduling?

Answer 1

Brooks says programmers are optimistic. Only in implementation is incompleteness and inconsistency apparent. He concludes that "running out" experiments are important disciplines.

The likelihood of failure or slippage is not null in every job. The probability that everything will go well is close to zero.

The cost depends on the workforce and capital, but there is no change! Therefore it is a deceptive and dangerous measure to use the man-month. These can only be interchanged if connections between tasks are not available.

• For a perfectly partitionable mission draw line down the top left to bottom right months (y) vs citizens (x).
• Display convex downward curve top left to bottom right for semi-partitionable function.
• Show flat line from the same months vs people graph for the unpartitionable mission.

For partitionable tasks involving communication, the completion cost must be applied to the communication:

• Structure.
• Intercommunication; if each pair is needed to communicate, "n(n-1)/2" increases costs.
• Meetings of > 2 people make it much worse!

The communications effort quickly dominates the effort to build a system. Adding further lengthens the timetable not.

Price of research still underestimated. Brooks recommends:

• Planning 1/3.
• Coding 1/6.
• Test of part 1/4, an early test of the system.
• 1/4 Process Check, all in-house parts.

TJP: Don't forget that writing test harnesses can be almost as much work as writing the actual code or at times more.

Delays are very demoralising during final testing.

However, Man and Mouth are interchangeable, online if the project consists of many individual activities and a situation in which team members do not have to interact with each other.

eg. Man             Months             Man-Month

        5 4                          20

       X2                       /2

       10                        2                          20

Let MMc be the average effort expended by a pair of people working on the earth, interacting with each other- defined to be earth dependent rather than N. Then:

-- Does not assume isolation of an original point. But communication among team members is complete.

MMn = Effort required without communication
MM = Total effort including communication:

Note: Task Force, Committee and Democratic Organizations need this. Not so for some.

Finding the optimum task staffing

-(Minimum time to complete):

Example: A 100 Man mouth task needs an average effort of 1 Man Mouth Communication per pair of project workers Nopt =?

With the task effort of 100 MM est., it wouldn't be a shocker if there were no communication considerations!