Wednesday, November 30, 2005

Distortion is rewarded, that's why it happens

Quite by accident I ran across an interesting article today by Edward Tufte. If you are not familiar with Tufte, he's probably the world's leading authority on how to present data for maximum impact with minimum distortion. I purchased a copy of his seminal work The Visual Display of Quantitative Information many years ago from a brick and mortar bookstore but you can conveniently buy it at Amazon.com.




Anyway, Tufte was engaged by the Congressional investigation into the explosion on re-entry of the space shuttle Columbia. In the article he is quite critical of the use by Boeing consultants of Microsoft PowerPoint to communicate the findings of their investigation into the inital collision on launch of a piece of insulating foam with the wing of the shuttle (which ultimately led to the disaster).

As bad as the Powerpoint presentation is, and all of his criticisms in the article are perfectly valid, I think he misses the point. PowerPoint is simply the medium. I don't beleive it caused the mis-communication, it simply facilitated it. (1)

I say this because I stumbed across the article after a day spent frustrated at a similar phenomenon (which not co-incidentally also has as its source a government agency), and it's not the first time I've had this type of experience.

Imagine trying, as I have been, to design a complex hardware and software system which must balance multiple competing objectives (cost with functionality with usability with security with compatibility and so on). A system which will cost millions of dollars, be used by thousands of employees and affect hundreds of thousands of the agency's clients and be distributed across hundreds of locations nationally and globally.

Now imagine being asked repeatedly, as I have been, to reduce your design to a single PowerPoint slide. What's more, imagine being asked to avoid using any of the standard notations that the software industry has developed to ensure semantic precision in its models, but instead to limit yourself to the standard slideware pallete of boxes and arrows.

In doing so you have to ensure that your slide communicates the complex design patterns that you have adopted to incorporate everything the software industry has learnt in the past 40 years. But you also have to keep it simple because the people selecting this multi-million dollar system don't have any knowledge relevant to evaluating your design.

Here's my point. In the end people sell what people buy. The Boeing consultants who created the PowerPoint slides that Tufte so rightly criticises did so because that's what their market demanded. The root cause of the problem is systemic and has to do not with deficiencies in the people doing the analysis, but in the processes which weed them out in favour of those who can spin a line to those too ignorant to know better, but unfortunately empowered to decide.

Had they wanted to use PowerPoint to communicate a clear message that forced people in positions of authority to make hard decisions they certainly could have. The problem is you only get to make such a presentation to politicians and their lackies once.


(1) Having said that, let me add this. Anyone who uses one of the standard PowerPoint outlines for a serious presentation is an idiot. I've used PowerPoint a thousand times, and I've yet to have a message to communicate that fitted one of those cheezy outlines. Perhaps when I start selling real estate (and hell freezes over) that will change, but until then create your own outline, one that logically follows the train of thought that led you to whatever conclusion you want to present.

No comments: