Sketches and Diagrams in Practice – Sebastian Baltes

More work today by Sebastian Baltes about who I tweeted this before:

This paper presents the rationale behind the SketchLink tool, as it researches how developers actually use sketches and diagrams. We know a bit about sketching already

2014-11-20 10.47.43

So, do we need another study? Yes, says Sebastian, because the other papers were small in scope in terms of only one company, one of area of interest or the number of participants. In this study, 3 people from 3 different companies were interviewed, from which 11 different dimensions of sketching arose.

After that, a questionnaire was sent out and spread through social and regular media. In that way 394 practitioners could be found to participate, consisting of mostly software developers and some software architects.


Sebastian investigated how common the creation and use of sketches was:

2014-11-20 10.52.57

Most of the sketches (60%) were made non-digital (paper or whiteboard) which is quite interesting given the fact that most people have a tablet and some offices have smart boards.

The time that people spend on sketches is short, 68% was created in less than an hour, and 74% was revisited at least once. These are all useful takeaways when building tool support for sketching.

Sebastian also looked into the envisioned lifespan. About 50% of the sketches was expected to have a lifespan of several weeks, which is curious, as most are created in a non-digital way. Would they save them with pictures of the whiteboard, maybe?

68% of the sketches were informal, little formal (UML) was used in them.

Sebastian was also interested in the level of abstraction of the sketches:

2014-11-20 10.58.32

Half of the participants said they thought that these sketches would help future developers to understand the source code, which is of course a nice result if you are building a tool like SketchLink 🙂

Finally, Sebastian looked into archiving practices (aha, my question) 58% of the sketches was archived. 94% of the digital sketches were archived, but also 40% of the analog sketches were preserved, which was more than expected, says Sebastian. Reason for saving it were: documentation, future use (to implement or to use as basis for other sketches), understanding or visualization. (note: I am not really sure what the difference between visualization and understanding is…)

Participants said things like: “It will be difficult to understand the code without the diagram”

Reasons not to keep sketches is that participants did not know how to save non-digital ones. Nice quote on why you would not archive is:

“I did not know how to combine code with the diagram, so I did not keep it”

All points in the direction that SketchLink is useful, especially given that documentation is often poorly written and out of date. Open questions remain, of course, like what distinguishes helpful from non-helpful ones.

The paper and the dataset are here: