Last week, we concluded a three-part series on Edward Tufte’s theory of graphical excellence. I enjoyed the series, but was provoked to read that Professor Tufte dislikes project charts (aka Gantt charts) because they have “regressed to Microsoft mediocrity.” I cannot let his claim go unchallenged, and I wonder if he would change his opinion if he were to examine what I present below. So, at some risk of beating a dead horse, here is what amounts to Part Four in the three-part series, but with my anti-mediocrity perspective.
The dictionary gives one meaning of ‘mediocre’ as
not satisfactory; poor; inferior: Mediocre construction makes that building dangerous.
I understand why Tufte might have judged many simple project charts to be inferior under his principles of graphical excellence. But I also think he’s overlooking how people currently customize project charts to make them good examples of these same principles. Consider the simple project chart shown below.
I have to admit that this chart is not very good. It is quite low in “data-ink” — Tufte’s word for the pixels used to convey project data, rather than to make the chart “look like a chart.” In particular, the green bars have a vertical dimension that tells the audience nothing at all, so most of that green ink is definitely not data-ink! Following Tufte’s lead, I can increase the proportion of data-ink by replacing the thick, green Gantt bars with circles at the start dates and triangles at the end dates. I can also group tasks into swimlanes and move the task names nearer to the symbols. This gets closer to Tufte’s ideal, because it conveys more information with much less ink.
However, this theoretically superior chart will probably lose my audience, because it requires them to decipher my creative but unfamiliar use of symbols for start and end dates. Adding a legend to explain the triangles and circles would lead to other problems while still not really overcoming the unfamiliarity of the new scheme. The typical audience member seems to know instinctively how to make sense of start and end dates on a Gantt bar. By contrast, my experience has shown that most attempts to replace Gantt bars with something more creative leads to audiences who are trying to figure out how to read the graph, rather than focusing on the insights the graph is trying to convey.
But there is a third way that follows Tufte’s graphical principles in moderation and still leads to tuned-in audiences. In response to the tension between maximizing data-ink and producing a chart that is easy to grasp, many project managers stick with Gantt bars but use the “wasted” vertical space in each bar to convey other vital information. Consider the chart below:
This chart utilizes swimlanes as in the second chart, restores the Gantt bars, and now displays additional data in the vertical extent of each bar:
- The color of each Gantt bar identifies the responsible resource group.
- The length of yellow “thermometer bar” inside the Gantt bar shows the percentage of the task that has been completed.
- The red bars at the tops of the Gantt bars tell which tasks or subtasks of summary tasks fall on the critical path.
There are other creative uses of the vertical dimension that I have not shown here. For example, another author has encoded a fourth datum (the lateness of the task) in the heights of the Gantt bars.
Tufte’s broad-brush indictment of project charts (aka Gantt charts) as being inferior would probably have some validity if all charts looked like the basic “green-bar” chart at the top of this article. But they don’t. Tufte’s criticism overlooks the fact that modern project charts are multi-dimensional, showing much more than start dates, end dates and task names. And because you can use the vertical part of Gantt bars in intuitive ways to show resources, progress, risk, cost, and criticality, many people now produce project charts that are rich in data, adequately high in data-ink, and very audience-friendly.
By incorporating Tufte’s principles into the classic Gantt-bar format, these modern versions of the project chart are far from mediocre, at least in my opinion. If Tufte were to examine a representative collection of such charts, I wonder if he would now modify his strong indictment?