Procedure. Entry and Exit Conditions

Procedure, Entry & Exit Conditions for Agile Spec QC, Extreme Inspection

New Inspection and Review Process.

This is an example of a Procedure, Entry Conditions and Exit Conditions that can be used for a Extreme Inspection/Agile Spec QC process.

EI.P1: Team. The specification writer (‘writer’) finds one other person (called a Checker) to (help) carry out the QC (Quality Control) of their specification.

EI.P2: Time. a meeting time, with maximum duration 1.0 hour is agreed, no matter the size of the document. (if the Checker is experienced, they can in fact do their checking at any time, alone, and report their results to the writer.)

EI.P3: Process-Brief. The writer makes sure the checker is knowledgeable about the following:

Readership: the spec's intended readership and their uses of the spec.

Rules: the specification Rules that apply (and their practical interpretation)

Major: The definition of Major defect, and how to spot them

Purpose: the purpose of the Spec QC process ( to help the writer get to real exit-able level of defect density).

EI.P4: Page. The writer and the checker will each select the same one logical page ‘at random’ (300 Non-commentary words) sample to check. The writer is now performing the role of a ‘checker’ on their own work. They should agree that the page selected is representative of the quality of the rest of the document.

EI.P5: Checking-Style. Checking will be done individually (but maybe in same room)

EI.P6: Checking-Time. The initial checking time will be 10 minutes. If NO Major defects are found by either checker. The checking process will continue for another 30 minutes. Even if no further Majors are found.

EI.P7: Early-Exit. If any Major defect is found (and acknowledged by the writer as a real Major defect) in the first 10 minutes of checking, then this will be considered a sign that the spec contains many more major defects. The writer will consider whether they want to stop the QC process and improve the spec, or whether they want to continue for another 30 minutes to gather more Major defect cases (to better signal what they need to rewrite).

EI.P8: Estimation. At the end of the checking time, the writer (or the checker if they decide to take reporting responsibility) will calculate the estimated Majors/Page in the current document (using formulas or tools supplied) and will report (on a form or to a database) all time used and results (Majors found, Majors/page estimated, decision to Exit or not, etc.)

EI Entry Conditions

EI.E1: Experience

At least one of the participants has done a well conducted successful inspection once before, or been briefed by a competent practitioner, or will be guided through the process by a competent guide (ideally an expert in this process).

Rationale: people need to have some reasonable sense of how to do this process, otherwise it can become corrupted. We believe we can avoid formal training in the method, but we need some knowledge and experience of it in place.

EI.E2: Author

The specification writer sincerely believes that the defect level is low enough to exit. They have done personal checking against the rules themselves and find no defects.

Rationale: the writer should take the trouble to make sure the spec id as clean as possible before inspections. They should not misuse people and time to compensate for sloppy work.

EI.E3: Source

Exited copies of all source specifications are available.

Rationale: there is little point in checking consistency against highly polluted source specifications.

(example by using bad Business Requirements to check new System Requirements).

EI.E4: Toolkit

An updated ‘Inspection Toolkit’ (with specification Rules, Checklists (for learning to apply the rules in practice), Process descriptions, forms, electronic support, intended readership role information) is available and is understood by the participants.

Rationale: This tool kit is the real definition of the Inspection process. This really determines correct use of the method.

EI Exit Conditions

EI.X1: Defect Density Condition:

Estimated Major Defects remaining per page is less than 1 per 300 Non commentary words (initially until end 2003 10 Majors, to get a lenient start).

FORMULA FOR ESTIMATION:

Assume 33% effectiveness of the 2-checker checking-process.

Total Unique Majors acknowledged by writer, found in the sample logical page, times 3, gives a reasonable estimate of Majors/Page. This is before writer correction of known Majors.

Note: the effectiveness for a 3 checker group is slightly higher say about 40%. This figure needs to be determined by your own measurement.

OPTION: we might manage the exit level at an individual writer level to gradually motivate them to improve by about 50% (defect injection) less per iteration of the write and check cycle.

EI.X2: Writer Veto

The specification cannot exit if the spec writer wants more time to improve it.