Tuesday, June 21, 2016

Intuitive Engineering: a few thoughts

I had a conversation recently about different approaches to problem-solving and used the term "intuitive engineer" when describing my decision-making approach.  For me, this has always been a logical way to explain the methodology I follow when interacting with engineering (or even just generic life) challenges, but the surprised reaction I received at this term has gotten me thinking.

In my experience (limited to a trek through academia and a few years in industry), engineering decisions tend to come in two major flavors:  intuitive (instinctual) & data-driven (evidence-based).  Obviously, the best kind of decisions are made with a balance between the two, but I have observed that most engineers tend to rely on one of the two more heavily.

To be more clear about what I mean: by intuitive engineering I am referring not so much to a touchy-feely emotional decisions, but more to a decision-making process that is heavily informed by previous experience and academic knowledge.  It can oftentimes result from instinct/reflex rather than careful calculation, and can allow for quick turnaround with little hard evidence on the table.  This isn't to say that it's uninformed; rather, it relies on a deliberate understanding of past experience and careful extrapolation into novel problems.  For example (one drawn from my life with a one-year-old): we've recently started building towers and fort structures with the pillows and couch cushions.  My daughter loves it and runs with reckless abandon through the "tunnels".  She also enjoys diving onto the pillow piles and flopping against any structure I've created.  Regardless of how soft a collapsing structure may be, I still will make an effort to create pillow forts that are relatively kid-proof and stable.  Having spent years building pillow forts with my own siblings throughout childhood (combined with an engineer's need to always create more perfect solutions), I automatically fall back on certain structures and supports.  Knowing ahead of time which arrangements of cushions are most able to stand up to rough-housing makes it trivial to build one just when my daughter needs to be distracted from grabbing mommy's phone.

Two observations about intuitive engineering that make increasingly more sense to me the more I speak with experienced engineers and scientists: the first is that intuition can also be hard to break.  We spend years in academia and even more in industry being told how the world works; so when a disruptive technology comes around and turns what we know upside-down, it can be a bit hard to accept the new paradigm.  Early adopters of new tech are not necessarily smarter or can better forecast industry changes: they just have an easier time reorganizing their internal tool belt to incorporate changes.  This is really where the system can break down: disruptive technologies and novel fields of development force changes in fundamental approach.  If we don't recognize the need to build new intuition and a supportive knowledge base, then the novel technology doesn't get implemented.  (More on this in a later post).  Secondly, intuitive engineering is ineffective is there isn't a basis of trust established between workers and their management.  I personally have had many occasions sitting in meetings where stating my gut-instinct in response to a hypothetical was dismissed for lack of substantiated evidence.  --Didn't matter that I was right (or that I would later provide the data to back-up my claims); until I established a record of accuracy with my peers, the trust wasn't there.  The onus was then on me to prove my engineering worth.  

On the other hand, the data-driven engineer can be equally challenging, even when relying on hard evidence.  Heavy on due diligence, this route can get muddled with disproving low-hanging fruit questions (aka "checking off boxes") rather than tackling the main problems.  In it's favor, however, is that it is much more effective in larger groups with disparate opinions.  At some point, data-driven decisions are less contestable... that is, assuming that the interpretation is sound.

The ideal is, of course, to find the sweet spot where you can balance the two types of approaches and their accompanying quirks.  Knowing how to establish the kind of trust that allows for leeway when making judgement calls before the data is in is immensely useful.  Additionally, creating a flexible "tool belt" that can adapt to new technologies, that can adjust on the fly, and that can be accessed quickly in new situations, makes a truly effective engineer.