Random Walks

How well does the almost universal Agile practice of “build it and see if they come” serve us (as developers, as customers)?

I suggest it’s time to rethink our belief that customers (and developers, for the most part) “don’t know what they want until they see it”.

My late, great colleague and friend Grant Rule used to refer to the practice, common in the Agile domain, of building (a portion of) something to see if the customer likes it as “random walks through the problems-solution space”.

Quality Demands Requirements

Philip Crosby, a widely acclaimed “guru” of Quality Management, defined quality as “conformance to requirements”. As simple and blunt as that.

Recently, I’ve been reflecting on my experiences with software product development, especially the development of “quality” products that customers love. In Javelin, we place special emphasis on de-risking delivery through explicitly defining the customers and their respective requirements. Not big-bang, up-front stylee, but incrementally, just enough each couple of days to build a little more of the product and deliver it to the customer(s) for their delight, confidence, and feedback.

But in our approach, requirements (in the frame of the Antimatter Principle we call these needs) precedes building anything. Agile shops these days seems to major in building something before discussing requirements (if they ever get discussed at all). BDD offers an exception, but how many shops do BDD?

Aside: In Javelin, we identify all stakeholders (a.k.a. all the Folks That Matter), discuss their needs (“Stakeholders’ Needs”, in Javelin parlance) and quantify them (a la Gilb – see: Competitive Engineering) in the form of Quantified Quality Objectives. Although:

• This all generally proceeds incrementally, rather than in a big batch up front.
• The information is always to hand by the time someone gets around to building the relevant part of the thing in question.
• The requirements come from dialogue(s) with the relevant Folks That Matter.
• The requirements need not get written down (documented) unless there are some Folks That Matter that need them to be.

People work from the requirements. Always.

Random Walks are not Our Bag

Random walks are not our bag.

By cleaving to the belief that customers “don’t know what they want until they see it”, and structuring the whole approach to development around this belief, Agile shops have no incentive to improve the way they work with customers to understand their needs. No incentive to improve requirements elicitation and capture. No incentive – or means – to prevent defects and deliver zero-defects quality. Indeed, this belief and its associated practices blocks us from working to continually find better ways to create useful requirements (formal statements of folks’ needs) from which to drive quality (cf Crosby) and the improving of relationships with each other (developers, ops) and with customers.

Is this emphasis on working-from-clearly-stated-and-agreed-requirements better? Well, in my experience it makes for happier customers, happier developers, and more successful products. I’ll leave it to you to decide whether and how that’s “better”.

– Bob