Talk:Software metric

From Wikipedia, the free encyclopedia

 This article is within the scope of Computing WikiProject, an attempt to build a comprehensive and detailed guide to computers and computing. If you would like to participate, you can edit the article attached to this page, or visit the project page, where you can join the project and/or contribute to the discussion.
??? This article has not yet received a rating on the quality scale.
??? This article has not yet received an rating on the importance scale.

Contents

[edit] SLOC confusion

I got here from SLOC, where I found a subtle confusion about, or at least a failure to explicitly distinguish between, SLOC as a software metric and as an indirect means of comparing programmer skill (a worker metric?). I see that the same problem exists in this article. Perhaps because the same problem is found in the use of software metrics? It seems to me it would be worth at least drawing an explicit distinction and stating that the two goals are not the same, but are related.

Effort of producing a given program is a separate issue from of comparing different programs (or program fragments or code snippets), one or more of which possibly being imaginary, which perform comparable tasks. Given two such competing programs, there is an assumption that the skill of the programmers who produced said programs can be determined by reasoning in reverse. The argument leads to the conclusion that software metrics can themselves be used to compare the quality different programmers. However the premise that we have at hand two roughly equivalent programs to compare for "efficiency" or otherwise has not been sufficiently examined, and thus the conclusion seems suspect.

In short, software metrics are not a means to evaluate workers. Software programs can be evaluated individually using metrics, which depend on a quantitative scale and can be determined absolutely. That they can then be compared is secondary. However, worker evaluation is not relevant except when comparing two or more workers, and thus is always relative. Confusing the two ideas makes both efforts less reliable.

[edit] Rules of Thumb Section Removed

I could not find any data backing up the Rules of Thumb section and apparently neither could anyone else . Section removed for questionable factual accuracy...

[edit] Rules of Thumb factual question

Wondering where the numbers came from for the Rules of Thumb section since they seem dubious, especially the 1,500 LoC figure for the Government programmers.

IMO, we need to either add references to support the numbers or remove the section entirely

Wulv

[edit] Schools of thought

Granted, it's been a decade+ since I took a graduate class in software metrics, but I left it with the understanding that there's two different schools of thought:

  • Rigerous: Studying metrics to develop a mathematical methodology. After developing a basis, then one could draw conclusion(s), and
  • Process: Metrics that analyse relationships between aspects of the code (e.g. LOC, # of defects, cyclomatic complexity, CMM stuff)

It's touched upon in the reference to process metrics.

Is the war over? Was there a winner?

[edit] Vocabulary

Is it wise to be using the expression software metric on its own? I believe that the word metric is not to be found in the CMM model and some researchers suggest it should be avoided. There is ambiguity in the different definitions of the term metric. Would the expression software measurement or at least a link to it be appropriate?

[edit] Criticism

The criticism section seems to me to be slightly off-target; it doesn't talk about the limitations of using metrics to judge software, but the related question of the limitations of using metrics to judge people and their work. It's a legitimate discussion, but I submit that it belongs in an article about management or psychology.

  • I think the criticisms must also be sourced. Who says its unethical? How inaccurate?BillGosset 02:01, 11 June 2007 (UTC)

[edit] Metrics vs Measurement

I haven't read the whole article, but the list of "metrics" is wrong from what my coursework has taught me. Example: Measure - Number of Lines of Code. Metric - Number of Defects / 1000 Lines of Code.

The definitions given to us in the latest class are as follows

  1. "Measure - provides a quantitative indication of the size of some product or process attribute"
  2. "Measurement - is the act of obtaining a measure"
  3. "Metric - is a quantitative measure of the degree to which a system,component or process possesses a given attribute"

CloudedIce 20:58, 30 October 2006 (UTC)

[edit] Science v. engineering

This is a software engineering topic, not a computer science topic. Why does the intro refer to computer science? Am I missing something? Adrian Sampson 23:37, 22 April 2007 (UTC)

[edit] Useful Articles on Software Measurement

I find the following articles in the CrossTalk February 2001 are good and worth reading:

  1. Software Measurement Programs and Industry Leadership by Capers Jones of Software Productivity Research Inc.
  2. The Nine-Step Metrics Program by Timothy K. Perkins of Software Technology Support Center.

They provide good analysis and classification of measurements and tangible guidelines to implement measurement. Follows the following link to read the articles:

Francis Law (talk) 04:03, 4 January 2008 (UTC)

dont u hav nay material regaurding the defination —Preceding unsigned comment added by 210.212.172.178 (talk) 05:05, 10 January 2008 (UTC)

[edit] Original Research or Editorialism?

Hey all. I'm concerned about this page being largely editorial in nature, or even original research. There are few citations, and a lot of opinion given. If this is common opinion in the industry, we should be able to verify it. But even the section on balancing metrics seems highly suspicious to me, and would be widely disputed by much of the current discussion within the Lean and Agile software communities. --Christian Edward Gruber (talk) 21:04, 13 May 2008 (UTC)

Additionally, the concepts range from software metrics for evaluating software development process control to metrics to measure personal developer productivity, but without a coherent framework for differentiating kinds of metrics, etc. --Christian Edward Gruber (talk) 21:06, 13 May 2008 (UTC)