Thoughts on Software Estimation
Alan Mark Davis · 2015
The average software project completes its job between 10% and 75% later than was originally predicted. The challenge in software schedule estimation is to convert some estimate of the “size” of a software product, along with some project characteristics, into an accurate prediction of when the software will be delivered. This chapter explores some common issues relating to the practicalities of usage of metrics and measurement, and offers some recommended solutions. The reasons for delayed delivery are combination of poor estimates and requirements creep as high as 2% per month. The chapter explores some of the problems we face in the 21st century concerning estimation, and recommends solutions to change the abysmal record of achievement to date. The problems include (i) the thinking that there is a “magic” measure of size and/or complexity; and (ii) developers play many games when providing estimates.