Essay

Knowledge Management and Decision Making 11.2022

Accessing the data around us

Information is a critical resource

In order to promote the importance of reading and education amongst young Americans, in 1989 NBC began airing a series of public service announcements under the slogan “The more you know” .
The campaign, which still runs today, has evolved with the addition of new themes over the years, but the core message remains the same: knowledge is power, and cultivating knowledge is one of the most fundamental aspects of life.

People learn in different ways, and there are different methods to acquire knowledge . While they employ different tools and techniques, these different teaching styles are designed to achieve the same outcome: extracting information and storing notions in an organized manner, so that they can be retrieved at a later time.

Today the world’s information is readily available anywhere and at anytime, to anyone who has access to the internet. But while ensuring fair access to information is a fundamental pillar to building an equitable society, just over 63% of the world’s population has access to the internet in 2022 .
Despite the potential to accessing information in real time has never been greater, there are contexts where institutional censorship and technological limitations (both at the individual or the infrastructural level) compromise how easily people can connect with data. To address the digital divide, the United Nations Educational, Scientific and Cultural Organization (UNESCO) has declared all access to public information a human right , and established the International Day for the Universal Access to Information on September 28th.
Alongside institutions, individuals alike have create initiatives to promote fair access to the internet. Most notably, Sir Tim Berners-Lee (inventor of the Internet) founded the World Wide Web Foundation with the goal of helping everyone connect to the web, and advocating for open web practices.

Forming judgement

Managing information

Daily and for the past two decades, digitalized countries have been generating such an incredible amount of data , that managing it all has become a challenge of its own.

Over this period of time each one of us has learned to parse high volumes of information on a daily basis by filtering, organizing and recalling each notion based on our individual needs. In his book “Thinking Fast and Slow”, Daniel Kahneman introduces some helpful concepts in framing how individuals form their judgement: slow and fast thinking refer to the ability of our brains to employ different type of processing capabilities depending on the type of information which is presented to us. Humans have an instinctual response to stimuli which we are trained to parse quickly (e.g. comparing a large object against a small one), while our slow thinking ability is capable of solving complex problems through sustained focus and effort.
Slow and fast thinking work constantly in support of one another: fast thinking becomes faster over time by consolidating information which we believe as true. The same data is then assimilated by our slow thinking system, which can either validate or dispute our initial judgement in a loop which reinforces our knowledge and understanding of the world around us.

The information lifecycle

Kahneman’s research exemplifies how knowledge follows a lifecycle of its own.
In the first step of this process, the reach of information describes under what circumstances the data is presented to us, factoring signals like the authority of the source and the delivery mechanism.

Once System 1 determines the validity of the information, it is passed to System 2 for further validation during the assimilation phase. In this state, we consolidate the information as part of our own sets of beliefs, and we categorize its use based on future needs: in that, we don’t learn directly from information, but mostly by reflecting and internalizing that data into actionable learnings for the future.

The last phase of the lifecycle is the practice stage: when our beliefs are put into action, we witness their outcomes within a specific context, and are able to validate their truthfulness through empirical evidence. In this stage, information reaches the peak of its epistemic value, and we are more likely to rely on this highly contextualized data, whose outcomes we have participated in. Information that is validated throughout this process acquires increasing value at each phase of the lifecycle, while data which is confuted at any point of the cycle loses relevancy. From that perspective, we can distinguish between learned and experienced information by the amount of bandwidth that they provide in the decision making process.

Utilizing information

While information tends to increase its epistemic value over time, there are inherit limits to both the amount of content we can effectively parse, as well as the speed at which data evolves through the information lifecycle.
Studies have demonstrated that more data does not always provide incremental value, and that right-sizing the amount of information we parse can help avoid analysis paralysis, one of the most common pitfalls of the decision making process .

In his book “Blink: The Power of Thinking Without Thinking”, Malcolm Gladwell explains how too much information can produce diminishing returns in our decision making process, and confuse us to the point of making suboptimal decisions.
Gladwell recounts that the amount of lives saved amongst patients admitted at hospitals for chest pains increased dramatically when ER doctors moved from evaluating all the physiological markers of a patient, to evaluating only three key indicators. What is remarkable is that the doctor’s diagnostic success rate was not affected, and they were able to diagnose patients much more quickly than before.

Now that the link between access to information, how data is metabolized, and the quality of our decisions making process is clearer, we can also think of our capability of making successful deliberations in relation with the time required to reach that decision.
Intuitively, many of us correlate decision making with risk management, and with the goal of reducing the likelihood of less favorable outcome for the decision maker. While risk is what we emotionally react to in the context of making an important decision, it is only a symptom generated by the lack of clarity around a specific context. From this perspective, the goal of the decision making process becomes to reduce ambiguity as much as possible , so that risk is contained, while chances of a favorable outcome are maximized.

In reality, we rarely have all the data needed to be certain that the decision we are about to make is unequivocally the right one. We are used to making choices by leveraging the partial point of view of the information available to us at that time, because even if we could know all the knowable, that decision would take an incredibly long time to deliberate on.
We compensate our knowledge gaps by leaning into our instincts and by using approximation techniques. As mentioned earlier, our fast System 1 responses become better over time as they get either validated or disputed by our slow System 2 reasoning, highlighting how making decisions with velocity helps us become better in the decision making process itself. We also tend to engage our slow thinking system and lean on earned knowledge, which has matured through the practice stage, whenever the risk of failure assigned to a decision is particularly high.
The Italo-American physicist Enrico Fermi was known for his ability to generate accurate estimates with little to no information available .
During the Trinity Test, a detonation exercise conducted as part of the Manhattan Project, Fermi was able to estimate the detonation equivalent of the explosive device in TNT by observing how a few pieces of paper he had dropped from his hand were moved by the blast. In many cases with decision making, imperfect estimation that is ballpark-accurate is more useful than a precise estimation which proves itself to be wrong.

Two distinct psychological profiles can be identified from how people approach ambiguity. Individuals who strive for a perfect deliberation and outcome are known as maximizers . This group spends a significant amount of time evaluating possible outcomes and carefully analyzing the impact of each decision by engaging their slow systems thinking to analyze all the data available to them.
On the contrary, the second group called satisficers tends to make a decision much quicker, and utilizing less information. Satisficers can make decisions faster as they let their System 1 thinking support part of their decision making process. The expectations of this group also tend to be more attainable, which supports the discovery that although maximizers invest a greater amount of time in deliberating their decisions, satisficers are more likely to be content with the outcome of their deliberations, as they are able to dispel ambiguity quicker and are less invested in the decision process making by itself .

Making decisions

We face thousands of decisions everyday. Not all of these deliberations are important ones, and we breeze through most of them with little to no effort. Psychomotor decisions usually deliberations we can make autonomously: we don’t pause to coordinate our movements when we are crossing the street, or when we are walking up a flight of stairs. We are also able to make most decisions that deal with our personal preferences very quickly: we can tell if we like a particular flavor, or a specific color almost immediately. Deliberations of this kind appear mechanical in nature, but they are the result of a series of impulses, reactions and connections that happen instantly in our brains, kicking off a chain of events which triggers muscles and sets in motion the actions initiated by our minds. We are extremely efficient at making low-stake decisions at high volumes throughout our day, but if we’re all expert decision makers, we are not all great at making decisions?

The impressive body of research conducted by Laurie R. Santos and  Alexandra G. Rosati compares the choices of primates to the ones that humans make under similar circumstances, to highlight how both species are biologically wired to prefer a certain set of outcomes compared to others, and in doing that, they share a similar set of biases .

Studies amongst humans and primates have demonstrated how both species tend to seek abundance over scarcity, and have uncovered they are both affected by the principle of loss-aversion, a cognitive bias by which losing something is more emotionally powerful than gaining an equal amount of the same thing. Similarly, we see both groups overvalue the importance of objects they don’t current have, compared to the ones they are currently in possess of, by a bias called endowment effect.
All of these phenomenons can become powerful influencers in both our fast and slow thinking processes, and ultimately affect our decision making.

Decision types

A decision is a deliberation that is made within a specific context, which produces an outcome (or a series of outcomes) that can be impacted by the decision maker. In order for us to be able to qualitatively speak about a decision, we must first be able to identify the possible outcomes.

By this definition, some prerequisites must exists in order for a decision to be made:

• Non-null: the deliberation must effect an outcome. If the decision does not determine or influence a specific outcome, the decision will have no impact.
• Non-equal: multiple qualitatively different outcomes must be possible. If the possible outcome is only one, or it’s qualitatively equivalent regardless of the direction we move forward in, a decision is not necessary.
• Greater-than: within the range of possible outcomes, we need to be able to identify a preferable set of circumstances. While in some cases outcomes are not binary and live on a spectrum of desirability, establishing what success looks like for the decision maker is of paramount importance in this process. If all outcomes are equally desirable we can also default to not making a decision.

As mentioned earlier, we are accustomed to grouping decisions by the impact of their possible outcomes. We think of a “big decision” as something that has a large impact on our lives, for example moving to a new city, or purchasing a car. While perceived impact is an emotionally resonant lens to use in evaluating the relative weight of a decision to the surrounding context, it is difficult to quantify unilaterally (e.g.: the impact of our best friend moving to a new city is different depending on whose perspective we analyze its effects from, ours or the friend’s). In order to build a standardized scorecard, we can analyze additional attributes of a decision, which can give us a more precise way to weight its consequences.

Lifespan

The first attribute is the lifespan of a decision, or in other words: how often do we get to make this decision?
A decision’s lifespan is usually correlated with the longevity of its impact, and therefore its consequences. The longer the timespan, the more effort we tend to put in our decision making process.
Let’s take the common example of a car purchase: car owners in the US hold onto their vehicles for an average of 12 years which statistically makes this purchase one of the longest-lasting decisions in an individual’s life. Since most of us understand the longevity of this decision, we tend to dedicate a greater amount of time and effort to researching, comparing and test-driving vehicles prior to deciding on one, and data shows that buyers spend an average of 89 days educating themselves prior to driving off the lot in a new car.

Reversibility

A second characteristic of decisions is the permanence of their effect. Purchasing a car is a decision with an uncommonly long lifespan, but also an easy one to reverse.
If at any point you were no longer satisfied or in need of a vehicle, you could sell it or trade it for another car with very little effort. Most decisions we face in our lifetime are reversible and can be made even in suboptimal conditions. Amazon’s management philosophy introduces the “two way door” metaphor for any reversible decision, while irreversible decisions are thought of as a “one way door” .

From that perspective, we are incentivized in utilizing a satisficer mindset whenever making two way door decisions in order to achieve an outcome as quickly as possible, while one way doors decisions require a longer deliberation period, typical of maximizers. Examples of this framework in action can be found in many companies building software at scale.
For example, as the team responsible for the UX frameworks of the Windows platform, the Windows Design Systems had to honor a multitude of one-way door decisions when modernizing the WinUI APIs during the release of Windows 11.

In fact, platform APIs have to be maintained virtually indefinitely to ensure backwards compatibility, meaning that once a specific taxonomy is decided upon it cannot be updated without introducing breaking changes. A notorious example of this phenomenon is brought to us by the early history of the web, when the HTTP referrer API was spelled as referer by the original developer almost twenty years ago, and has been unchanged since.

Defying determinism

The journey towards improving our own judgement is full of hidden surprises. As we strive to fight determinism to procure an optimal outcome for ourselves, humankind is able to look at ambiguity as a driver of innovation and ingenuity.

Epistemic uncertainty is one of the biggest driver of societal evolution, and history has proven times and times over how critical thinking and deliberate decision making has impacted how we evolve as a species. The development of Git, JavaScript and Linux are examples of how conceptual integrity, Pareto inspired product development, combined with the ability to make fast and accurate decisions are fundamental in producing impactful and lasting work.

As we embark on our individual journeys, it is most successful to focus on outcomes that we can impact directly, while maintaining awareness that each one of us will encounter many known (and unknown) unknowns along the road, and that experimentation and learning will contribute to better decision making in the future. This perspective reminds us that as humans we are both fallable and resilient, and that the pressure we put on ourselves on achieving a specific outcome has to be balanced with the ability to forgive oneself for all the suboptimal decisions we will make along the way.

References

1. All Companies Should Live by the Jeff Bezos 70 Percent Rule, June 27, 2020. ↩︎
2. Rasmussen University, 4 Types of Learning Styles: How to Accommodate a Diverse Group of Students, August 17, 2020. ↩︎
3. Statista, Number of internet and social media users worldwide as of July 2022, July, 2022. ↩︎
4. UNESCO, Access to Information Laws. ↩︎
5. World Wide Web Foundation, Establishing the open Web as a basic right and a public good. ↩︎
6. Seed Scientific, How Much Data Is Created Every Day? +27 Staggering Stats, October 28, 2021. ↩︎
7. PubMed, When knowledge is demotivating: subjective knowledge and choice overload, 2014. ↩︎
8. Frontiers in Neuroscience, Ambiguity aversion in rhesus macaques, September 17, 2010. ↩︎
9. Wikipedia, Fermi problem. ↩︎
10. Psychologist World, Maximizers vs Satisficers: Who Makes Better Decisions?. ↩︎
11. The Wall Street Journal, How You Make Decisions Says a Lot About How Happy You Are, October 6, 2014. ↩︎
12. National Library of Medicine, The Evolutionary Roots of Human Decision Making, June 15, 2015. ↩︎
13. Wall Street Journal, Americans Are Keeping Their Cars Longer, as Vehicle Age Hits 12 Years, June 14, 2021. ↩︎
14. Autolist, How Long Does it Take to Buy a Car?, May 12, 2021. ↩︎
15. Farnam Street, Reversible and Irreversible Decisions. ↩︎
16. Autolist, Amazon Founder Jeff Bezos: This Is How Successful People Make Such Smart Decisions, November 3, 2022. ↩︎
17. Inc., All Companies Should Live by the Jeff Bezos 70 Percent Rule, June 27, 2020. ↩︎
18. Wikipedia, HTTP referer. ↩︎
19. Opensource.com, How JavaScript became a serious programming language, October 28, 2020. ↩︎
20. Wikipedia, History of Linux. ↩︎
21. Wikipedia, There are known knowns. ↩︎