Had another heated discussion with a self-professed agile Guru who declined the offer to take the conversation public and place his arguments here, so I will do it on his behalf, but allow him the anonymity he desires. I will refer to him as John for argument’s sake and leave out some of the less important stuff.
The argument went like this:
John: Your perception of agile is not right, you just don’t see the real benefits of agile because you are not a true agile practitioner, you just use it as a looser form of waterfall when it suits you, but you never really took the faith. We are delivering year in and year out in a way we never could with a traditional approach.
Ed: I’m inclined to take that as a compliment. I hold deep suspicion of anyone who takes a methodology too seriously above and beyond the actual delivery of an end. I have indeed used agile to great effect, but wouldn’t use it except in situations where I believe it is the better approach.
John: That’s just it, you believe you know better, but if you tried it in this other situations you would find, like we do that it delivers more every time.
Ed: I would be interested to see the definition of delivers for the sake of comparison. I won’t discount the possibility that you are right, because like you, I don’t have facts and figures with which to make a real argument, but I do have compelling reasons for my decisions that I am happy to share with you.
John: And they are?
1. Given an ideal world, the shortest and cheapest way to an end result in software is to know exactly what you want to achieve in advance, then know exactly how is best to deliver it via software; have experienced people to do it, test rigorously against your acceptance criteria and roll it out once working perfectly. I doubt anyone would argue with that. I do accept that it probably has never happened and possibly never will, but it is the perfect scenario.
If we both started together and you used agile while I used a waterfall cycle, there’s little doubt that I would finish first and come in cheaper, though given the smoothness of the task, you would also perform well.
I would expect in that scenario, to get the business design right first time and agreed, to get the system design right first time and to have a very low level of defects in testing and instant acceptance by the client. Your trial-and-error approach would waste precious time and produce unnecessary artefacts.
2. The reason a business and I include in this definition the majority of public bodies, invests in IT is to save money or earn money. A commercial operation needs a return higher than the cost of capital as a fundamental criterion and it must also qualify as best use of limited capital.
The toughest part of a business case is nailing down the cost. If you don’t know the cost with a level of confidence, you can’t estimate the return confidently and you don’t have a case.
The agile argument about “fit for purpose at maximum cost” works well provided “fit for purpose can be relied on to deliver the expected returns and assuming that it can be achieved, but ultimately, unless you can define the requirement well enough in advance to estimate the cost within reasonable boundaries and prove a business case, you are unlikely to ever get off the ground and probably should not.
I don’t believe there would be a great deal of difference between the two performances is that case.
Anyhow, that’s all well and good, but there are many other scenarios where the business case is proven by the do nothing scenario alone E.G. defence, environment etc where the costs cannot be calculated accurately, but there is no argument about whether it should be done other than to figure out what will work best.
Ed: I agree with this entirely and in any of these many situations, I too would use an agile approach to remove the likelihood of a colossal error by learning as we go along.
It went on further, but the point has already been made
This conversation went on a little longer and covered more ground, but I do feel that a lot was revealed about the importance of using the right approach for the right situation.
I do believe that a good business analyst with strong consulting skills can still earn his keep time and again by getting the requirements defined , understood , tested, verified and agreed at the outset so that accurate costing and planning can occur and the most cost efficient route can be taken.
I often see an agile approach taken as a substitute for consulting skills rather than as a good strategic decision and I do believe that this does nobody any favours least of all the agile approach itself.