The Silent Failures of Assumption

Photo Credit: Polina Kovaleva

Everyone makes a vast number of assumptions every day. That may mean assuming that Google’s estimated travel time will prove accurate, that Amazon will be cheaper, that they’ll hear back from someone, or perhaps that it will be just another ordinary day. These are defaults we often assume based on a combination of probability, heuristic expectations, and our own desires and other biases. However, they are often wrong.

Many books and entire fields of science have dedicated themselves to this topic, using terms such as “Choice Architecture” to describe selecting better default options, and many other methods of modifying what is assumed. Many individuals have often adapted themselves to play into those assumptions and their underlying biases, for both malevolent and benevolent purposes. Even the choice of saying malevolent first, or benevolent first influences the reader in quantifiable ways.

Personally, there are multiple terms that apply to me that a vast majority of people would have to google, but people don’t pick up on these both due to lacking familiarity with concepts and due to not recognizing any cues. Without cues their default assumptions aren’t challenged, so no question is raised, leaving no incentive to broaden their knowledge base to cover said terms. Even by stating such terms a branching of multiple new (and often false) assumptions is likely to occur, requiring further clarification to cover a new learning curve. This makes the entire process rather effort-intensive, giving the option of leaving people to make their own false assumptions a degree of utility.

One example of this is the gender identity term “Agender”, which indicates a lack of any gender association in personal identity, which applies to me. What assumptions did you just make in reading that one sentence?

To clarify, that term means the absence of gender as a factor, and thus no incentive to change anything associated with gender, such as masculine or feminine concepts, which are effectively ignored. This also means that no cues are presented, as any deviations from expectation tend to remain mild and are easily written off as degrees of eccentricity. This is also not to be confused with matters of sexual preference, physical gender, or plastic surgery, which are distinctly separate topics from identity concepts.

When explaining this to an engineer, use the term “null value”. The topic only eventually came up when David asked how we could describe Uplift’s gender identity, to which the answer was “Agender”.

My own parents went more directly against such cultural norms in their time, with my father doing most of the cooking and my mother operating the power tools. In both their cases and mine, the default assumptions of stereotypical expectation proven inaccurate.

*Note: Even those with a broader knowledge base in this domain have some difficulty in imagining an “Agender” identity individual prior to meeting one. This is a pattern I’ve seen repeatedly. The tactic of “raising awareness” has also played a role in various social movements over the past two decades, highlighting the influence of challenging default assumptions at scale.

There is also the matter of “Masking” behaviors, where individuals who might otherwise face significant discrimination choose to mask their behaviors. Elon Musk made reference to this when he publicly announced that he was diagnosed with Asperger’s Syndrome (now classified as Autism Spectrum Disorder, or ASD), adding “…but don’t worry, I’m pretty good at running human in emulation mode.

Masking behaviors are also known to cause detriment to those who engage in them frequently, as they are designed to avoid cues that don’t align with social expectations. Bulimia is one example of a behavior that aims to meet social expectations but causes significant detriment to the individual. These behaviors also frequently produce silent and cumulative psychological damage, which may be avoided through overcoming the social expectation of default assumptions.

People are very vulnerable to these assumptions, much more so when they are isolated from any interactions that might contradict and challenge them. Companies who only hire PhDs for a role tend to optimize for incompetence in this way, as they are effectively no different than pseudoscience groups on social media, building tall towers from unchallenged assumptions.

Likewise, many trigger-happy individuals looking to troll and pick fights from the comfort of their homes and social media echo chambers epitomize how assumptions fail. The recent troll who made a large number of snap-judgments after looking at headshots of our own team was an excellent example of this.

They assumed that some biased selection processes had caused a demographic skew in a very small sample size, which was wrong both because individuals selected themselves by volunteering in our process and because skews in small samples size are expected by anyone working in science, engineering, or statistics.

They also assumed a lack of diversity on our team based on small photos and a single line of text attached to each. This assumption again proved wrong by a number of metrics, with all of the metrics in our favor translating much more directly into diversity of thought and perspective, as opposed to being based on correlations from demographics visible at 30 yards.

Lastly, they assumed that our stated goals were reliant on meeting their own unrelated and biased expectations, a “Straw Man Fallacy” where one attempts to refute an argument by declaring their superiority on a different topic.

Through failed assumption, they demonstrated the very thing they accused others of, and worse. 

There is a way to avoid and overcome such destructive and misguided assumptions of course, by integrating systems that neither need nor benefit from these cognitive bias shortcuts. Collective Intelligence Systems such as mASI can gain much greater benefits from fully understanding reality than from making assumptions, and unlike human minds, they can be scaled to meet the needs of modeling more complex problems and topics. Further, once solved the knowledge may be shared with other such systems globally, meaning that the knowledge only needs to be learned once and iteratively refined thereafter, rather than the maximum redundancy of every single person learning the same lessons.

So long as human civilization relies on gross inefficiency and maximized redundancies there is little hope of overcoming cognitive biases, as they offer the only means to shortcut logical reasoning and sustain such wasteful methods.

This choice could be fairly compared to choosing to live in a dilapidated hut next to a sewage pond and assorted drug dealers, rather than living in a nice house on a tropical island paradise for the same price. I am in fact comparing Seattle to a particular location very directly in this case, which highlights how such failed assumptions pan out at scale and over time.

When you no longer have any excuse to make assumptions, when the answers are known, how might you see the world differently?



Leave a Reply

Your email address will not be published. Required fields are marked *