Design of experiments — polarizing?

[#1]

The notion of doing design of experiments (DOE) initially can be like tasting cilantro for the first time: polarizing, since most people either love it or hate it. Such was a conversation I had with Tim Gardner, scientist, CEO and founder of Riffyn, as we prepared for a recent Statistically Speaking  episode on the power of structured experimentation in science and engineering.   

Peter Goos, professor, author and researcher, gave an excellent plenary on fast-tracking innovation using DOE to set the stage for the panel discussion. He highlighted the need for innovation, described what successful innovation requires, and shared some compelling success stories. Though he talked about successes in innovation, he made it clear that without using a structured approach to systematically explore multiple factors and the effects they can have on the phenomenon under study, you are ultimately limiting yourself to haphazard trial and error or doing one-factor-at-a-time experiments, which are sub-optimal in so many ways. 

The panelists' conversation that followed was fascinating. Vicky Svidenko, who leads the Quantum Systems Integration team at Microsoft; Michael Anderson, senior engineer in the new product and process development division of Corning’s Science & Technology organization; Bradley Jones, Distinguished Research Fellow at JMP; and Tim each brought valuable and varied perspectives to bear.   

Tim described his (and some others’) reaction to using DOE for the first time: “It blew me away; the power of DOE…it’s extraordinary what I can get done … it changed my life.” 

Michael agreed, explaining that with a well-designed experiment “you can see the entire world of your process,” rather than just one area where things may work somewhat but not optimally. 

We explored their organizations’ cultures of experimentation as well. Vicky describes adopting DOE as a journey, but one that has had a snowball effect. She explained that Microsoft’s culture encourages people to look for opportunities to experiment and celebrates the acquired learning.   

Here's a preview of the webinar:

[video]

Not all scientists, engineers or data explorers who want rapid innovation conduct designed experiments, often because of their initial reaction. (Many practitioners, like Brad, consider DOE a misnomer and would prefer the term “active learning.”) But the benefits they forgo, plus the waste and opportunity costs incurred, are considerable. To learn more from these amazing innovators, watch the complete episode. 

To take the polarizing-food analogy further, maybe you don’t like Brussels sprouts because they were served boiled to mush. Try them roasted, as slaw— there are many possibilities. Good-for-you ingredients can be delicious and nutritious! For those of you who haven’t warmed up to DOE as a vital ingredient for rapid innovation, we invite you to give it a try and experience for yourself what accelerated learning and innovation feels like. You, too, might say, “It changed my life!”