TheBabelExperiment

In the article "The Babel Experiment" [1], Vladimir L. Pavlov and Anton Yatsenko describe an experiment where students have to design a software system. But there is a catch: They are only allowed to communicate via UML, no verbal or written communication is allowed.

What I found interesting is the following summary:

"... we had two teams working on the same task. One team was limited to using only the UML ... The other was allowed to use the speech [i. e. spoken words] in addition ... [The first team's] diagrams were more detailed, more elaborated and elegant. ... When a task is discussed by team using the speech, everyone understands each other, so it seems to participants that there is no need to map 'obvious' things ..."

So, the members of group A were not allowed to speak which each other, they were restricted to using gestures and the diagrams. Group B however could use all means of communication; the result: the diagrams of group A were more detailed and accurate, as they also included things, group B regarded as obvious.

This made a lot of sense to me and I was wondering if one could use the idea for getting better requirements. What do you think?

[1] The Babel Experiment. Vladimir L. Pavlov, Anton Yatsenko. Proceddings of the 36th SIGCE Technical Symposium on Computer Science Education. Feb. 2005.

-- MarkusSchnell 2006-09-05 (edited 2006-09-06)

I missed something. I don't know how to intepret the word "speech." Help. SteveSmith 2006.09.05

Steve, I added a clarifying paragraph. -- Is it clarifying? -- MarkusSchnell 2006-09-06
Markus, Yes, it is. Thank you. SteveSmith 2006.09.17

I remember doing a simulation in a management class three decades ago that was very similar to what you describe. I see the experiment as showing the difference between a monologue and a dialogue. A dialogue enables the opportunity to test whether a shared meaning has emerged; a monologue prevents testing.

This made a lot of sense to me and I was wondering if one could use the idea for getting better requirements. What do you think?

Perhaps. Give the UML to the receivers. Allow the creators to observe but ask them to provide zero explaination. Have the receivers develop a list of conclusions based on the UML. For each conclusion, have the receivers list their assumptions. Ask the creators to comment about the receivers' conclusions and assumptions and whether they were what was intended.

SteveSmith 2006.09.17

That's a lot like the ambiguity measurement we use. - JerryWeinberg 2006.09.18

Updated: Monday, September 18, 2006