Open source reader

Tuesday, May 23, 2006

5 reasons why we fail software estimation ... but I have a solution for you...

Keith Pitty has an old entry titled Software Estimating and Economic Forecasting where he quotes the following:
  1. Optimism
  2. Over-Confidence
  3. The False Concensus Effect: the tendency to assume that others think like us
  4. The Curse of Knowledge: the tendency to assume that others have learnt from the same sources
  5. Status Quo: the tendency to assume that things will remain the same
#1 & #2 are almost the same, and they are killers.
#3 should not be a problem for senior projects managers
#4 & #5 are complex things ...

Today, he wrote an entry called Tom Looy says "Stop Estimating!"
Here is what tom says:
  1. Stop estimating
  2. Prioritise requirements
  3. Gather requirements just in time to enable developers to start coding
  4. Measure progress (only completed tasks count)
  5. Measure using smaller sized tasks
  6. Start measuring as soon as possible
  7. Use measurements to validate original estimates and adapt the plan to reality
Is it really doable in a real world scenario ? Come on ... It could be achieved only for internal projects only ...
I honestly think you need to get back to the old trick:

Internal project estimation * 3.14 = Project estimation !

Simple enough ! :)


  • Sometimes you cannot go without an estimate, typically when you are competing for a job.

    This estimate will be more or less off - and to me it seems that the actual skills don't matter that much, but the domain knowledge does.

    After doing a fairly detailed estimate on a project (we came out with 3500d) my boss told me that we shouldn't worry about the precision of our estimates - it won't matter, because your wrong estimates won't kill you, the things you didn't estimate do.

    In the end we both where right: He sold the project for 1500d to the customer (he compensated with some "license-fees" to claim that he was selling a "product").

    By Anonymous Carsten Saager, at 8:20 PM  

Post a Comment

<< Home