Sunday, February 26, 2006

Simple Requirements Example

The example below is for people with an analytical bent who are interested in understanding my definition of "requirement":
"A requirement states the least stringent condition that must hold to solve or avoid a problem that a prospective customer faces."
Assume we are developing a temperature control system, and the problems that prospective customers face include:
  1. Prospective customers feel too cold or too hot in their homes.
  2. Prospective customers will feel frustrated if it takes more than one minute of their time per day to maintain a comfortable temperature.
We would likely include in our requirements a Maintain Comfortable Temperature use case and attach an ease of use constraint to it.

Here are two possible ease of use constraints:

a. The system will be a thermostat with a dial to set the desired temperature, a switch that determines cooling or heating mode, and an on/off switch.
b. For a user that fits profile 'x', it should take no longer one minute of his time per day to maintain a comfortable temperature in his home.
It is conceivable that we could solve problem 2 with a temperature control system that did not have the user interface specified in constraint 'a'. Therefore, ease of use constraint 'a' is not a requirement, as it is not the least stringent condition that must hold to solve problem 2.

Constraint 'b', on the other hand, flatly restates the prospect problem in terms of a negative condition. It is therefore a requirement.

3 comments :

Scott Sehlhorst said...

Hey. I chuckled when I read this. Not only did you support your definition of a requirement, but you also supported my definition of a requirement.

Who knows - perhaps in all of the real world examples, we actually agree on this, even though we use very different language to describe our perspectives.

I completely agree that example "a" specifies the implementation and is therefore a solution-design artifact and not a requirement.

I also noticed that you had an ideation step in going to example "b" - you decided that the problem of regulating a 'target temperature' was the problem you were choosing to capture as a requirement for the system.

You could have chosen to buy the person a wardrobe of layered clothing that they could change throughout the day (that's what I do when my wife is in charge of the thermostat). You could have proscribed a program of temperature desensitization, so that the person had a broader definition of "comfortable". Or you could have used hypnosis to make the person unaware of anything short of life-threatening extremes.

Roger L. Cauvin said...

Well, I tried to be careful here. I'm not sure if what you call "ideation" is a good thing when it comes to requirements.

I framed one of the problems as the ease (in terms of time expended) of maintaining a comfortable temperature. Given this problem, there was no "ideation" in expressing the requirement as a limit on the amount of time it should take to maintain a comfortable temperature. The requirement did nothing more than restate the problem in terms of its inverse.

Perhaps the problem statement contains some implicit "ideation". To the extent that's the case, I think the problem statement (and thus the requirement) is flawed.

Don't get me wrong. "Ideation" sounds like a great thing. But it seems more like design than requirements to me.

Roger L. Cauvin said...

Rereading this entry and comments, I now don't think it's fair to contend that it contained any "ideation". By "maintain comfortable temperature", I did not mean setting a target temperature. Certainly, an A/C system with a thermostat might be a natural solution to the problem. But I explicitly rejected the alleged constraint containing that assumption as design. My intent was to frame the problem of feeling too hot or too cold as a requirement. That's what I meant by "temperature". Perhaps "perceived temperature" would have been clearer.