Sunday, August 16, 2015

Design Thinking is About Feedback

Deming PDCA Cycle
This image reminds me of a conversation I had with a co-worker who had just installed a large and very powerful electro-mechanical system; He was gingerly adjusting some control inputs and observing the results. I asked him if there were any sensors on the device to monitor the output. There weren't, which explained why he was being so careful with the joystick.

There are dozens of examples of this all around us everyday. Railroad locomotives. Commercial aircraft. The Tesla Model S. For some of you it's a kitchen appliance. What these have in common is that in "the wrong hands" they can do a lot of damage.

Control theory is an interdisciplinary branch of engineering and mathematics which deals with the control of dynamic systems and how their behavior is modified by feedback. The core idea behind Control Theory is that the behavior of a system can be regulated thru a process of measurement and adjustment.

Feedback Loop
In control theory, a system which does not monitor it's output is referred to as running "open loop." Such systems can create dramatic, unpredictable and costly results. Remember the stock market “flash crash” on May 6th, 2010 or the Tacoma Narrows Bridge? Start with a complicated and slightly unstable system, disturb it at the right moments and the next thing you know, its "out of control."

PDCA as a Controlled Design Process

As designers, we create systems that can be very dynamic, have many inputs and often produce unexpected results. Both our design process and it's outputs can benefit from the use of feedback.

This is what Design Thinking does. It utilizes user feedback to refine the output of the design process.

This raises an interesting question; How many times thru the cycle do we need to go before we're sure we have it right? Again, Control Theory may offer some insight;

Many systems have something called Natural Frequency. Its the period of oscillation the system settles to when it's disturbed. For bells, organ pipes and guitar strings, it's the frequency they resonate at when struck or plucked.  There are other natural frequencies related to the rotation of the earth, moon and stars. Natural Frequencies even exist at the atomic level. We also call it oscillation.

Another aspect of oscillating systems is that they run down. Over time, the strength of the vibrations falls off. This is called damping. Critical Damping is the rate which reduces the time to settle to zero amplitude most efficiently.  The chart above shows the result of four different rates; Over Damped, Critically Damped and two Underdamped.  The total number of swings across the time axis is a little over five.

Let's re-context this idea to a design effort.  The goal is to stop the oscillations in the shortest number of cycles - with the least wasted energy. The distance from the x-axis is how far away we are from a solution that meets the customer's needs.  The distance to the right of the Y axis is how much time it takes to converge on that solution. You can think of the numbers on the x-axis as periods of time; minutes, hours, days, years, the duration of your current project schedule.

If we define Critical Damping as having a perfect understanding of all the customer's needs at the beginning of the project (Features, Performance, Schedule, & Budget) we converge on the best solution with the least wasted effort and the shortest amount of time. It's optimally efficient.

Typically there are unverified assumptions and ambiguities at start of a project. We resolve these by asking questions, rapid prototyping and testing solutions on the stakeholders. During that process we're often somewhat off course, but continually correcting. The under or over damped paths take more time to traverse and use more resources in converging on the final solution. 

The Critically Damped curve is actually within 10% of the final goal in less than half the time. The underdamped paths do cross a 90% point sooner, but they still need further correction, consuming more resources to undo mistakes. The smallest area under the curve, (resources used) comes from the Critically Damped path. 

Here is another expression of this idea taken from Mattson and Sorensen's Fundamentals of Product Development;

This graph illustrates the percent of improvement in a prototype as the number of iterations increases. Model 1 gets a 20% imorovement. Model 2 improves it further by about about 37%. By Model 3 the rate of improvement begins to really fall off, but the solution is nearly 80% of the way to completion.

Interestingly enough, starting with an understanding of even half of the problem produces convergence in the same amount of time as a critically damped solution, but uses a bit more resources. This is how Design Thinking shortens schedules and reduces costs by using customer feedback to refine the solution more quickly and efficiently, a.k.a "Lean".

The analogy isn't perfect, but if we consider design to be a process of learning and correction with needs as inputs and solutions as outputs, it's instructive to realize that the time and effort needed to converge on a solution might be reduced by half over other methods. It also explains why empathically listening to the customer and acting on it is the most critical tool in the successful designers toolbox. 

No comments:

Post a Comment