DIRE Warnings, Part 1: Introduction
How Diversity, Inclusion, Representation, and Equity are supposed to work and where they go wrong.
Suppose you were to read tomorrow that a demographic analysis of careers shows that the percentage of Amish in the military, in STEM fields, and in business leadership were exactly proportional to their percentage of the population. Should we celebrate that we’d finally achieved equity for the Amish? Or is it more correctly described as achieving true diversity in these fields? Or is it inclusion we’d achieved? Maybe representation? These terms can certainly be confusing. Or, would such a situation be evidence we’d become an oppressive, assimilated mono-culture that disrespects diversity?
What if I told you that everything you knew about diversity, inclusion, representation, and equity (DIRE) is wrong. In this series of articles, titled “DIRE Warnings”, I aim to show you how the principles of DIRE are supposed to work. I compile the DIRE principles across multiple sources and shows how they all point to a single common design and set of definitions. These sources include training material, research literature, legislation, psychology, and engineering design principles. (As a Canadian, I will be referring to Canadian legislation, but there may be equivalent legislation elsewhere.). The articles will also address common misconceptions and poor design that result in failure modes of DIRE efforts.
The fundamental issues resulting from these reviews include:
Diversity is often confused with representation. Diversity is about the individual and the variety of unique dimensions, qualities and characteristics we each possess. Diversity is not about differences in immutable traits such as race, ethnicity, or gender, or more generally about a single dimension.
Representation is often incorrectly applied to general population or as quotas. Representation refers to the numerical presence of specific characteristics with respect to the availability pool for a given occupation and region. It is a method of checking for barriers, not a generalized target outcome.
Equity, diversity, and inclusion (EDI) are often misapplied to drive categorical outcomes. EDI is about treatment of individuals via internal processes and culture regardless of categorical statistics.
EDI can change potentially change demographic statistics, but it can go in either direction depending on relative prevalence of biases that filter people out versus coerce assimilated mono-cultural conformity.
EDI should not focus on one-dimensional traits, especially not surface-level immutable traits like race and gender, and not by training or lecturing. At best this causes interpersonal conflict and decreased performance; at worst it evokes group psychology that rapidly degenerates into a toxic work environment.
EDI should focus on informational diversity, intellectual debate, and the removal of biases and barriers in processes. Let the demographics work out organically as a result.
Target statistical outcomes should not be used, such as proportionality based on general public demographics. Such targets drive EDI to failure, including:
a) Disrespect for diversity. Proportionality assumes no diversity and indicates an assimilated mono-culture, not a diverse multi-culture. It assumes the dominant values are held by all.
b) Over-constraining the system. If y = f(x), given x is a diverse multi-culture input and f() is a fair system respecting EDI, then the outputs y must be free variables. Constraining outputs will either (i) fail by halting the whole system (e.g., no hiring), (ii) fail a constraint (e.g., skills, qualifications), or (iii) burst components at their weakest point (e.g., under/over-staffing).
c) Undercutting representation, such as coercing self-identity higher than the availability pool to do an end-run around the statistics to hit targets without changing anything.
d) Short-cutting representation, such as hiring directly based on target traits to achieve targets by violating human rights codes without addressing why numbers were low in the first place.
This series of articles will take you through each step of the design of DIRE systems to demonstrate these conclusions, and finish off with examples for what applications of equity, diversity, and inclusion are supposed to look like. I hope you will join me.