Talk:Agile software development

From Wikipedia, the free encyclopedia
Jump to: navigation, search

Agile impact on quality section would be nice[edit]

Just going a Google here and there one can find Agile advocates (a.k.a. religious proselytizers) suggesting that Agile software development processes do not adversely impact quality whereas one can also find Agile critics (a.k.a. David Hume-class Skeptics, capital "S") note that there is no body of metrics or suitable research which suggests that Agile processes (1) provide any value and (2) impact quality either way, beneficial or detrimental.

One of the problems I see just doing Google Scholar checks is that pale studies -- informal "at a glance" studies -- in to whether Agile processes contribute anything beneficial to software development are behind pay walls, the same goes for such Google Scholar-found publications which discuss Agile's impact on quality. All such studies are absolutely not scientific in any way.

For the extant article, it would be nice if there was a section that pulled together even the minimal research conducted in to whether Agile processes contribute anything or impact quality in any way. I would like to propose adding such a section complete with references and citations however linking to articles behind pay walls is prohibited, and the legitimacy of any such reference -- even to IEEE or the Journal for Computing Machinery -- are going to be biased.

What do you think? Damotclese (talk) 22:07, 7 June 2017 (UTC)

This article already contains a section for Criticism. If, as you suggest, you have evidence that agile approaches do not improve outcomes, or even that they make them worse, then I would suggest adding a sub-section to the criticism section, and exploring the ideas there. Be sure to include references, though, as anything critical that is unreferenced is sure to be reverted pretty quickly. I look forward to reviewing your contribution.
Incidentally, while you are right that you should not link directly to works behind a paywall, you can still cite those works -- much as you would for a book or other publication you would have to buy in physical form. They are all fair for citation and reference. Davidjcmorris  Talk  06:03, 12 June 2017 (UTC)
I'm not suggesting that Agile processes make quality worse. I'm noting that there is no science-based research that says that Agile works to improve quality or makes quality worse. Did you bother to read what I said? The question is whether the extant article should have verbiage noting that there is no science for this concept. Damotclese (talk) 14:55, 12 June 2017 (UTC)
Thank you, I did read what you wrote and, I think, resolved the concerns you voiced vis-a-vis articles that are behind a paywall, the ones that you felt you could not cite (whereas in fact you can). I welcomed your contribution and, I believe, was also quite supportive in terms of how to go about it. I did not intend my response to suggest your point to be purely about negative impact, although as I re-read it now I can see that is an interpretation of what I wrote. It would perhaps have been better written as, "if you have evidence that agile approaches neither improve outcomes nor make them worse then ...".
However, as you have highlighted your argument, let me pose a question for you: How are you be able to determine that there is no science available? If you have not accessed those articles behind a paywall, how are you able to determine that they are all pale studies (a rather pejorative term) and not scientific in any way?
To phrase this in a more scientific way: You cannot prove or disprove a null hypothesis without evidence. If you truly believe that there are no articles available that describe scientifically-based research into the impact of agile practices on quality, then you need to provide evidence for that viewpoint.
While you should not add content with unsubstantiated original opinion, you could instead add content that references others who claim there is no impact (as you did in your earlier comments), so long as you provide evidence that they said those things (in the form of citations). As I suggested, the appropriate place for this would be the criticism section. Davidjcmorris  Talk  02:01, 13 June 2017 (UTC)
That's just it, finding science-based research that cover impact on quality, research that is published on line, has been unsuccessful, Google Scholar points to non-scientific research involving asking polling questions from developers, and polls do not follow the Scientific Method. Point being that the extant article should note that there is no science which supports or detracts the ideologies of Agile process. Damotclese (talk) 14:10, 13 June 2017 (UTC)
Thanks. We should pause to consider the leap from an inability to find scientific research to saying that no scientific research exists. They do not equate. Some might consider it un-encyclopedic and un-scientific to say that they do, and making a statement that no scientific research exists is a pretty provocative statement to make when it cannot be evidenced. I think all we can really say is that you have so far been unable to find any scientific research. If you added that to the Criticism section, then if others know of some they can add references to it. This would not invalidate your initial assertion, that you have been unable to find any, and would move us on in a positive direction. Davidjcmorris  Talk  21:58, 13 June 2017 (UTC)
Separately, we could discuss research methodologies, as there are plenty of accepted qualitative approaches that construct hypothesis and proofs from interviews and surveys. This is perfectly acceptable, when it is done rigorously, especially in the social sciences. It is even better when this can be strengthened by quantitative evidence, which is probably the harder science you are after. I am confident that plenty of both types of research exist. I referenced some examples of each in my masters thesis on The Paradox of Agile Transformation: Why trying to be Agile stops organisations from becoming agile. I will look back to those, and other sources that I have to bring what light I can to your efforts. However, in the meantime, do not hesitate to add your comments to the Criticism section, as suggested multiple times above. Davidjcmorris  Talk  22:05, 13 June 2017 (UTC)
There is no scientific basis for Agile being better at quality or product development, no academic research has been done, there are no peer-reviewed journal entries that cover it. The OP noted that polls are conducted which solicit opinion, but that's not science. The article should probably note the fact but I expect that believers in Agile would revert such updates since Agile is a religion to believers in it. BiologistBabe (talk) 14:58, 14 June 2017 (UTC)
Two assertions without evidence.
  1. No academic research has been done? Just a cursory glance at the public domain Google scholar, reveals over 200,000 entries on agile quality research. Some of those entries will be of questionable quality, for sure ... Google scholar does not have peer-reviewed checks on what it includes ... but I can see entries that link through to academic papers and peer-reviewed articles on the impact of agile practices on quality. I also know that running the same search behind a university paywall would generate many more entries that would be more academic. If you put such an assertion into the article without references, then for sure it will be reverted, because personal opinion is not encyclopedic. However, if you can evidence that there is no scientific research, then please do add it to the criticism section. If you add an evidenced statement and it is reverted without warrant, then I give you my personal undertaking, for what that is worth, to protect its integrity until it can be refuted with evidence that there has been scientific research, academic papers, and peer-reviewed articles.
  2. Agile is a religion? This is a tired cliché. While there are agilistas (the bane of my working day if I encounter one), there are far more pragmatic agile practitioners who recognize that the right approach depends on the context (and yes, a more planned up-front approach is appropriate at times). For example, check out the rigorously researched Cynefin framework from professor Dave Snowden. Davidjcmorris  Talk  18:03, 14 June 2017 (UTC)
Yes, exactly, Agile ideologies are not covered in peer-review journals even after 15 years, that's why there is no evidence that it works, no evidence that it does not work. That's why the extant article should have commentary about the lack of scientific validity. But as we can see, there is, as noted previously, a religious aspect to the belief that it improves product and quality, ergo we couls expect believers to not accept the lack of science... As we see in the talk page here. Damotclese (talk) 13:47, 15 June 2017 (UTC)
Maybe you can help me understand your viewpoint. I am someone who has lived through these changes. I see the benefits of a more planned approach for some types of work and a more exploratory approach for others. That is far from a religious belief, that is a pragmatic engineer's understanding that it is horses for courses. I started my working life in the 1980s as an IMB/370 Assembler programmer and then COBOL. Good times. The only prevailing methodology at the time was the staged approach, which is mostly referred to as waterfall these days. When our cycles were long, when technology seemed fixed, and when we thought we could lock down requirements at the outset, this seemed fine. After all it was the only thing we knew. It was that or follow no process at all. That too was OK for small systems but not for large complicated systems. Then ... new technology kept coming at us in wave after wave, markets kept changing their needs year by year and then more frequently. As the mix of our markets and technology became more complex, we needed a way of being able to respond more rapidly to these changes. Just In Time development and Rapid Application Development looked promising, The Unified Process, that was designed to evolve, was good too. But the cycle times were still too long. Then we got DSDM and Scrum and XP, and these were dramatic improvements. All of these were evolutionary steps, they built on the strengths of what came before, they equipped teams to cope with different circumstances. They got labelled as agile in 2001 by some guys on a mountain, but again this was a group of engineers figuring out how to build on what they knew. For some reason, every 5 years or so, as more and more organizations adopt more agile ways of working, it is described as all new. This is scary for some people. Then a lot of training firms, consultancy firms, and software toolkit providers got on the bandwagon, and of course their marketing engines took over. Four legs good, two legs bad. Everything that came before was evil, everything you do with the new way was good. Then they started in-fighting. To this old-hand, that was the most astonishing stage. To see the founders of Scrum denigrate the founders of SAFe was disappointing. To then see the founders of Scrum fall out and form their own separate associations was heart-rending. But I digress. Yes, for sure, there is a lot of heat and noise, fuss and nonsense written and spoken about agile practices. But those people do not represent the mainstream. They do not speak for me. Agile practices are no silver bullet. They will not make things magically better overnight. In fact, what many organizations struggle with is that adopting agile practices will expose organizational dysfunction, and that can be scary. And, more importantly, to this old pragmatic engineer, they are not appropriate at all times. I even wrote a blog on the research that waterfall was more effective than agile on very short projects. Can you please help me see where anything in this can be viewed as religious? Davidjcmorris  Talk  18:41, 15 June 2017 (UTC)
To move us on (if that is now possible). I am incredibly supportive of you adding this criticism into the article. I have pointed out the best section for it to go into. I have described how to cite material that is behind a paywall. My only advice to you was that your assertion should be referenced. That is sound encyclopedic practice. However, rather than just publish, you seem content here to keep repeating that there is no research, that nothing has been published in any peer-reviewed journals over the last 15 years. Unless you have evidence or references to corroborate that, all you can really do is express the opinion that you have not been able to find any yet rather than claim that none exists. Again that is the scientific way of describing your position (see my earlier point on the challenge of disproving a null hypothesis with no data). I would love to see that published, so that we can then see whether anyone in the community is able to find some (if it exists ... I think it does, but I cannot prove that yet). To help this, and avoid people just throwing out what you have described as pale research, perhaps you could take a moment to define more precisely what would you consider to be rigorous scientific research and what types of journals you would you consider to be valid and worthy. Davidjcmorris  Talk  18:50, 15 June 2017 (UTC)
Re-reading the original question, I returned to the main article and scrutinised it for assertions that agile practices have an impact on quality. There is no religious assertion that it has any impact. There is one reference to a claim in 2003 that agile practices improve quality, although no assertion is made that the claim is valid; it may be assumed by the reader that an absence of commentary on the claim leaves it standing as if it were commonly held to be true. There are also two somewhat hidden references to surveys in which people self-reported improvements. In light of this, I fail to see the level of passion expressed at the article. However, as stated above several times, the question poser is welcome to add a paragraph to the criticism section; so long as it is able to reference someone of renown who has this view and is able to point to where in this article any claims are made. Davidjcmorris  Talk  01:23, 18 June 2017 (UTC)
That is correct, there are no scientific studies that show that any of the so-called 'Agile' processes actually work, but that's the same for a lot of things that are basically religious beliefs which are marketed for profit. A good example is 'Common Core,' a set of religious beliefs that are sold to governmental educational entities under the non-scientific claims that it some how improves education and, ironically, reduces cost.
There is no evidence that Agile does anything useful or, for that matter, anything harmful. It is religion. The article should note that fact that there are as yet no science behind the claims Agile makes. TrainsOnTime (talk) 14:59, 29 June 2017 (UTC)
Thank you for adding your weight to the discussion. As neither of the previous contributors have yet been able to provide an adequate definition of what they meant by scientific studies, perhaps you would be willing to share you wisdom in this regard. If we can establish what the criteria are and what methodologies you would deem to be scientific enough, then we can embark on a reasoned and logical search and discuss the results of that. In best encyclopaedic etiquette of course. Let us please keep the baseless mud-slinging out of it. Thank you. Davidjcmorris  Talk  07:01, 30 June 2017 (UTC)

New lede section[edit]

I've re-written the lede; it still needs expansion in a few areas. Specifically, agile anti-patterns should be mentioned, or else that entire section should be moved to a sub-page. Also, something regarding "agile v. waterfall" might be added.

I don't think anything so far is controversial, but the changes in total are fairly large. Power~enwiki (talk) 23:07, 14 July 2017 (UTC)