Normal?

Math and science, like music, are universal languages. They enable us to encapsulate and communicate meaning. They also work well as metaphors and analogies. Examples like 1+1=2 for "Obvious" and E=mC2 for "Powerfully Simple" are many.

One powerful example from the fascinating world of statistics is the famous Gaussian Distribution (Named after mathematician Carl Friedrich Gauss) also referred to as the Bell Curve for its inverted bell shape when plotted. Many mathematicians and statisticians contributed to the work around continuous probability distributions like Pierre-Simon Laplace who proved the CLT (central limit theorem) reinforcing the Gaussian Distribution. The Bell curve gained popularity above all other distributions because of how frequently we see measurement results of many phenomena (like people's heights, exam scores, lifetime of lightbulbs, milk production of cows! and many more) land nicely distributed along the inverted bell curve with most readings in the middle and the numbers fading as readings stray from the middle majority.

"Normal"

The Gaussian Distribution (a.k.a. The Error Curve given it first became famous with Gauss modelling errors in astronomical observations) was named the "Normal Distribution" in the 1890's (Ironically the distribution that started as the error distribution was named "Normal" by error and attempts to correct it only gave it more fame!). This name implied that measurements landing in the middle of the bell curve were the "Normal" measurements.

The Logic Gaps Problem

Like all languages, Math can be used to describe reality or equally to distort it! The Gaussian distribution works perfectly well when we feed it a phenomenon we can measure well and represent in limited scores. When we do, it provides us with symmetry about the "Zero Error" line of measurements. As it became popular as the "Normal Distribution", it got irresponsibly used to define what a normal person is. This was a big logic jump with nothing to support it. What I think of as the Logic Gaps problem starts when we try to do one of two shortcuts (or "logic jumps"):

• Oversimplification: When faced with complexities our science cannot harness yet, we resort to reducing the complex reality down to simpler constructs we can communicate (and calculate/compute). This Reductionism can lead to misrepresentation that when fed to credible science spits out incredibly misleading "Findings". Famous examples are Exam scores (Measuring Academic and work potential), IQ scores (Measuring intelligence!), Employee Performance scores (Measuring career potential) and more. When these complex concepts (Some of which we don't fully understand like Intelligence) are forced into a limited scored representation (Basically, dumbed down!) to match our existing scientific and computational abilities, they lose most of their meaning and context. Many will rightfully debate the same problematic pattern extends to using amazing modern technologies like Deep Learning in AI to harness superior complexities of Languages where we insist on reducing language to mere repetitive patterns, so it fits within our DL capabilities!
• Forcing Reality into the curve: Depending on the company you work for or the education system you know; you may find it "Normal" that a max 20% of employees will receive outstanding results each year or a max 2% of students will get an A+ in a given exam cycle. The same employee delivering the same performance in another year, or another department will most likely receive other results and the same student answering the same percentage of correct answers in another exam cycle will most likely get another grade! The root cause of this paradox is the skewed points of view. Employees and students view this as a fairness problem (i.e. If I'm an employee delivering outstanding performance against my role expectations, I should receive outstanding results irrespective of context. The same goes for a student answering +97% correct answers and expecting an A+ irrespective of context). On the other hand, Employers (who still use this system) and education systems view this as a budgeting problem (i.e. Awards are a function of performance results and award budgets are fixed early in the financial cycle (Not a function of latent actual performance) and Exam results are the regulating function of how many will get into which universities and programs). Hence, to differentiate performance, there can only be 20% outstanding and to manage university capacity, there can only be a specific number of high-grade students. The "Normal Distribution" will still hold true given even within outstanding performances, there are some that are more outstanding than the rest and that can be a good enough numerical basis to answer fairness concerns.

At the intersection of Oversimplification and Forcing Reality into a bell curve is the root cause of many confusing, harmful, and even dangerous examples. The one that I believe did more harm in the business world is the bell curve adoption in Employee performance systems. It inevitably leads to an individualistic culture that can tear down otherwise amazing teams from within,

One of my proud moments in Microsoft was back in 2013 when we announced, "No more Curve" and "No more Ratings". It was the start of clear signal that Teamwork and Collaboration is the way to succeed and that focusing on what matters with a deeper understanding of impact is the way forward. Fast forward to today and "Normal" is clearly redefined as: Simultaneously Deliver results, contribute to others' success, and build on theirs.

In closing

They say we measure what we treasure, and this is true. It's only when we short-cut and expect molded results to support pre-established opinions packaged into theories, that we end up "surprised" when one after the other breaks the mold and performs outside of "Normal" expected results. If we're not open to learning and adjusting, biases from the unconscious to the conscious and in all flavors from "Prove it Again" to "Tightrope" become inevitably "Normal"!

P.S. on the extreme side of this topic is bending science to work against the norms of diversity, Racial justice and more and use it as a justification to legitimize unthinkable actions. Left unchecked in the wrong hands, the same pattern can, and unfortunately does! get even uglier with ideologies like Eugenics oversimplifying human quality and trying to force it into a curve and proceeding to justify/legitimize decisions on who is feeble-minded and who gets to reproduce! Many people involved in these practices believed they were following solid science. This is one more reason for us to get actively engaged in the debate and actions towards Responsible use of Innovation. Today, the powerful rapid cadence of innovation can be equally powerfully good or powerfully bad. We have to decide! A fitting example is the work we get to do in Microsoft to support the advancement of Reponsible AI.