Browse Prior Art Database

Method for an uncertainty principle of software release determinism

IP.com Disclosure Number: IPCOM000011850D
Publication Date: 2003-Mar-19
Document File: 3 page(s) / 34K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a method for improving the prediction of software development completion dates. Benefits include better setting of customer and management expectations for delivery of software and enhanced project management methods.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 53% of the total text.

Method for an uncertainty principle of software release determinism

Disclosed is a method for improving the prediction of software development completion dates.� Benefits include better setting of customer and management expectations for delivery of software and enhanced project management methods.

Background

        � � � � � The accuracy of predicting the date of software development completion increases exponentially as you approach the actual date of completion. Prediction approaches 100% accuracy as you reach 100% completion (see Figure 1). This principle is based on the use of historic data to predict future results. For example, it is common to use bug discovery trend data to predict when the software will become stable enough to declare it ready for customer/field use.� However, there are limits to this practice and this paper describes how to build these limits into your prediction.

Description

        � � � � � The disclosed method describes an ‘uncertainty principle’ of software completion determinism. The method includes the principle that a period of uncertainty occurs as a development project approaches 100% completion. Without this principle, project-completion prediction becomes random within a bounded margin of error, which is typically the standard deviation of the historic data.

        � � � � � Although historic data can be a very accurate predictor with large data sizes, the prediction becomes highly randomized with small data sizes. For example, by examining the amount of time required to fix hundreds of bugs, the average time to fix a bug can be accurately predicted. However, the standard deviation for fixing a single bug can be quite large. Thus, when using this data to predict the software’s stability timeline, as the project approaches one standard deviation of the historical data, predicting the actua...