Monday, June 13, 2005

New Color Modeling Group

Following a long discussion thread in my Yahoo! group, where I realized that there is a broad ignorance and misunderstanding of the role color modeling plays in FDD, I have created a 2nd Yahoo! group, this time to facilitate discussion of The Coad Method, domain modeling in color technique. While people would like to learn more about this powerful domain modeling and architecture method, there really isn’t  a place to do it. Finally, I’ve solved that problem. There is now a place to discuss and learn modeling in color, class archetypes and the domain neutral component. If your interested in this area and how it contributes to agile methods and long term prdoctivity in software development then I would encourage you to sign up and join in. It’s been 6 years since Peter Coad published the last version of his method for analysis and design. Since then we’ve proved the value of it again and again. In my recent post detailing FDD benchmark metrics, I didn’t mention that I attribute up to 50% of the best performance to proper execution of The Coad Method. If you’re not modeling in color, you could be leaving a 2x performance improvement on the table. So join up and contribute as we all learn how to model for agility at http://groups.yahoo.com/group/colormodeling/

Click to join colormodeling

[Update:  There are over 40 members already who are actively learning how to model domains using color archetypes and the domain neutral component pattern. Daniel Vacanti and I are leading the members through a full worked example - payroll.com, a fictional outsource payroll vendor. If you’d like to see this technique in action join the group and follow along as we build a domain model a little bit every day.]

If you don’t know what color modeling is, or simply want an online reference to all the best material, then I’ve compiled this handy list

The book chapter that started it all, Java Modeling in Color Chapter 1, by Peter Coad

Stephen Palmer’s Coad Letter collection archived at his own site

And some papers from my own site, including one by Daniel Vacanti…

Arguments about Color by Daniel S. Vacanti

and the rest from me…

Advanced Domain Modeling (BorCon 2004)
Coarse-grained Components from a Color Model
Whole Part Relationships in the DNC
Coloring with Demeter
A History of Color Modeling
The Case for Class Ownership
Architecture Control Board
Quality Assurance and Over-modeling
FDD and Legacy Code

Sunday, June 05, 2005

Approx 44 Mins

“Our flight time from take off until landing this evening will be approximately forty four minutes.”

I was sitting in a small single isled regional jet. Outside it was marked with the insignia of Delta Connection, a loose affiliation of aircraft operators who provide the feeder services to Delta airlines. We were on the tarmac at Cincinnati, Ohio bound for Pittsburgh, Pennsylvania. Our flight attendant was Sylvia. She was about 35 years old, with a Grace Jones meets Dennis Rodman hairstyle of short, crisply cut blonde afro curls. It looked pretty good on her - in a sexy but no nonsense kind of a way. It was her job to insure our safety for the next three quarters of a hour, whilst stuffing us with peanuts and ginger ale. It was dark outside the windows already. The 20 passengers sprinkled around the cabin were all keen to get to Pittsburgh and find their way to their respective beds, whether they be at home or in some homogenous business hotel - like the one I was destined for.

So what is the point? The flight was going to take three quarters of a hour. Or according to Sylvia, “approximately forty four minutes.”

Now hang on, we humanoids have 5 fingers on each hand. We tend to approximate in 5 or tens or in multiples or halves. Three quarters is a good approximation. We don’t approximate in multiples of 11 like 44. So what gives?

Sylvia knew how to subtlety set expectations. She understood variation. By offering us approximately 44 minutes, she was setting the expectation that the flight time was scheduled as 44 minutes but that it may vary within some range, say 40 to 50 minutes. That would be normal and we wouldn’t be “late”. In fact the total scheduled time from push back to docking was longer and indeed we were to be early arriving in Pittsburgh even though the flight took 45 minutes.

So what is the lesson here?

Note the subtlety in language. It’s variation aware language. If she’d said “our scheduled flight time tonight is 44 minutes” then we might have reasonably felt aggrieved when we landed after 45 minutes. Approximately 44 minutes is the language of variation. It’s the language of a Deming style agreement that conforms to a mid point and spread. “scheduled for 44 minutes” is conformance to plan and specification language. It sets the expectation that exceeding 44 minutes is out with specification and represents non-conformant quality. “Scheduled for 44 minutes” makes for unhappy customers, “approximately 44 minutes” sets the appropriate expectation and makes for happy customers.

WSA QA SIG

Here are my slides from the WSA QA SIG meeting last month. They represent an update and reworking of the presentation from the Seattle chapter of the ASQ last September. I’m getting beyond bullet points with these presentations. They are more and more just a set of talking points. So they are pretty hard to follow unless you were there to hear the words. I’ve been asked to write a paper on these slides which will be published in a management journal. That won’t happen for about 6 months. When the paper is published I’ll put it here too so that you can read the words that go with the pictures.

Thanks to everyone who came along to the meeting. You were a great audience with lots of good questions. I hope I didn’t disappoint you.

Download Understanding Variation in Software Engineering [PDF 3.7 MBs]