[kplato] PERT Variability

[Jim Sabatke]

Hi all,

First, congrat's for getting out a first version!!!

When we first started discussing this project, we spent a lot of time on 
PERT statistics.  In my opinion, PERT charts are a waste of a LOT of 
paper, but the statistics are extremely useful and separate a mediocre 
planner from a truly useful one.  They are also easy to understand and 
calculate.  They also (usually) give much more accurate resulting 
delivery times than hard durations.

Another very valuable tool is the charting of certain project variables. 
  The most useful:

1. Budgeted Cost of Work Scheduled vs. Actual Cost of Work Performed. 
This gives the schedule variance.  You can show this to your manager or 
team and disclose if the they are behind or ahead of schedule/cost.  You 
all probably realize that in software projects especially, that often 
tasks are done out of order for various reasons.  Therefore, GANTT 
charts are often deceptive.  When building a bridge, concrete HAS to be 
poured and cured before starting steelwork; but in software the tasks 
are often just a collection of things that need to be done, often in no 
particular order.

2. Actual Cost of Work Performed vs. Actual Cost of Work Scheduled. 
This is the cost variance.  Are we delivering this project at the 
defined cost?

Attention to those two graphs will give notice of disasters while there 
is still time to correct them.  They also give 'paperwork' to managers 
that they can easily understand when they ask if the project is on time 
and within budget.  Managers don't want to hear about task variability 
as a rule, so these, plus GANTT charts, tend to placate them.

As one of the originating 'thinkers' on this project, I am very happy to 
see it moving forward.

Regards,

Jim