Summary (TL;DR): The perfect information heuristic is a state of mind where you assume you have access to all available information. It allows you to more specifically and narrowly define what decision you are trying to make and allows you to look for data that, at first blush, may seem impossible to get. It is effective in counteracting the availability heuristic. Its shortcomings are that it is more work, many times what is available is what matters, and it can give you overconfidence in the decision you ultimately make.

Lastly, one could argue that this “heuristic” is just sound decision making and doesn’t really need to be elaborated on. I don’t have a great rebuttal to that, but clearly, I think it’s worth discussing.

On the Perfect Information Heuristic…

What is it: There is a quote that I am very fond of by Laplace:

“We may regard the present state of the universe as the effect of its past and the cause of its future. An intellect which at any given moment knew all of the forces that animate nature and the mutual positions of the beings that compose it, if this intellect were vast enough to submit the data to analysis, could condense into a single formula the movement of the greatest bodies of the universe and that of the lightest atom; for such an intellect, nothing could be uncertain and the future just like the past would be present before its eyes.”

— Marquis Pierre Simon de Laplace

If the premises are valid then I think the logic is correct. Thus, if one had access to all available information in the universe, one could see the future and therefore make the most optimal decision. While that is impossible, it is still a good heuristic to apply to decision making, especially in countering the availability bias. To more specifically define the perfect information heuristic: it is a state of mind where you assume you have access to all available information, as such, what variables would you examine in making the best decision.

Application:

Using the perfect information heuristic for decision making would start as follows:

Pretend as though you have perfect information, specifically define what you are deciding on and determine what variables are the most salient to the decision. …hmm, seems like a “duh” moment. However, the availability bias(1) creeps in due to: specifically defining decisions are nearly impossible, most decisions you make you have some level of information already and, additionally, decision making is circular, as you learn more you redefine what you are looking for and as you redefine you conform to what information is available.

Let’s say you are thinking of going to college (an imprecisely defined decision – for most, college is a means to an end – define what your “ends” are). You have heard over the years that college educated people make way more than non-college educated individuals (people have initial information). So, what happens is you imperfectly define your decision, should I go to school, then you list variables that you know exist: one being do college educated people make more than non-college educated.  As you start looking for this information you stumble on all the careers college educated individuals have access to vs non-college educated – this information is available and should be incorporated (decisions are circular). By the end, you confirmed that college educated people make way more than non-college educated and you realized that there are more career opportunities for college educated individuals. So, you may not know what you want to do, but if you go to college you will make more money and have more career options once you ultimately decide on what you want to do. The data supports it and you can safely say you made a deliberate, informed decision.

So, let’s apply the perfect information heuristic to deciding on going to school and why the above is flawed.

Define the decision specifically: Do / should I go to school Perfect information would dictate that you know exactly what you want: I want to be an accountant making $75,000 a year, working at xyz place, living in abc location, etc. Are you going to capture everything, no, but starting with the premise that you could know exactly what you want changes the way you think about the problem. So, you meet halfway:

Imprecisely defined decision	Modified Decision (still flawed but better)	Perfect Information (tons of variables still unaccounted for)
Do / should I go to school?	Living at xyz geography is most important and making $>50k a year (which allows me to have a good-sized house and the ability to go to Florida twice a year is important.) Will school help me get there?	What is the best way for me to become an accountant making $75,000 a year (so I can do xyz with the money), working at xyz place, living in abc location, etc.

 

Already you can reframe the question from one that society (you need to decide if you are going to college and the answer should be yes) impresses (read: makes available to you) upon you to one that has more relevance to your life. However, let’s say the modified decision you are making is only “I want to make >$50k a year, should I go to school?” In that slight reframing, you now look for ways to make >$50k. From that, you find a stat that most jobs where people make >$50k the individuals are college educated and it lists the careers. You then apply the heuristic to that stat to see what additional information you can get. Start with perfect information and work backward.

Direct, perfect information question: Would I be able to get one of those jobs? (Unanswerable)

Indirect questions, pretending you have access to near perfect information: If they went to college what degree did they get?  What companies do they work for? What skills are they performing? Where do they live? How old are they? How long did it take them to get to >$50k? etc. etc. Just throw questions on a page until you can’t think of anything more.

Pair down to importance: Of the above questions, only a few will hold real influence/matter to the decision you are making. Let’s just pick degree and geography as being important factors to you. (after asking those questions you realized you don’t want to be a science major and you don’t want to live in a big city).

Find proxies: The stat listed only showed the careers of college degree holders. It didn’t show the degrees or where those individuals live, so you must find a way to get that information or approximate for that information. Let’s say you trolled LinkedIn and sampled 80 individuals. Of that sample, 65 were science majors and/or lived in a big city. That leaves 15 individuals and maybe three professions that met your criteria.

Re-analyze: Initial stat was way more people with a college education make >$50k a year. Revised, more relevant stat: I would, effectively, be eligible for three professions that require a college education that makes >$50k a year. If I didn’t go to college, I would be eligible for these other five professions that make >$50k a year. Compare those specifically then.

Shortcomings:

It is more work. If the decision you are making doesn’t appear to have much impact on your life then why expend the energy to make a more optimal decision and, in some cases, get to the exact same answer. And in some, if not many cases, you will get to the same answer. There is wisdom embedded in what society impresses upon people. Questioning for the sake of questioning or being contrarian for the sake of being contrarian isn’t in and of itself valuable. Lastly, in going through this analysis, you may lose the humility that knowing every possible outcome is impossible. You can be wrong and you will be wrong. When you are wrong, be flexible and change. By applying this heuristic the goal is to improve the ratio of good to bad decisions, not to eliminate all bad decisions.

Conclusion: See the summary at the top of the page…

(1) The availability bias, in short, means that individuals will only leverage the information that is immediately available to them regardless of its value. People act on what they can see or measure, even if what they are measuring holds no value. I think this manifests itself in both deliberate and indeliberate decisions. For this post, I will focus on how the perfect information heuristic can help in deliberate decisions where the availability bias rears its head.

 

 

(please read “What are Mental Models” to understand the emphasis on “MY” and why I include useful)

Unfortunately, nothing can be 100% accurate / verified. Also, in the Soros writing, “According to Popper, scientific laws are hypothetical in character; they cannot be verified, but they can be falsified by empirical testing… One failed test is sufficient to falsify a theory, but no amount of confirming instances is sufficient to verify it.” However, while nothing can be 100% verified, there is a spectrum of accuracy that I believe roughly follows academic disciplines. When I am thinking through mental models, I categorize them in these broad buckets and following subcategories. This provides a rough understanding around the level of accuracy of a mental model I am thinking through. (Physical sciences are generally more accurate than social sciences with social sciences generally being more accurate than humanities)

Is this an all-encompassing list? Not even close. Models will be added and they may be subtracted. The below provides an educational framework that will hopefully allow me (and you) to understand how the world works. My goal is to build out a database of essays around these models that I can refer back to. Most of what I have listed I have read about and have some knowledge and thoughts on. The next step is to transcribe those thoughts and continue learning new models.

A few notes:

• There is a unity to science that isn’t appreciated and I don’t think explored enough. As you traverse from the physical sciences to the humanities they are effectively all linked and become more and more complex. This complexity creates the opportunities for more errors and is why I have grouped them in the aforementioned categories. However, just know that when I am exploring group think or sunk costs there will be much more that could be explored that isn’t feasible to do so in an essay. To fully appreciate the increased complexity I will quote a book description (yes…a book description of a book that hasn’t been released at the time of this writing). From Robert Sapolsky’s “Behave”: the first category of explanation is the neurobiological one. What goes on in a person’s brain a second before the behavior happens? Then he pulls out to a slightly larger field of vision, a little earlier in time: What sight, sound, or smell triggers the nervous system to produce that behavior? And then, what hormones act hours to days earlier to change how responsive that individual is to the stimuli which trigger the nervous system? By now, he has increased our field of vision so that we are thinking about neurobiology and the sensory world of our environment and endocrinology in trying to explain what happened. Sapolsky keeps going–next to what features of the environment affected that person’s brain, and then back to the childhood of the individual, and then to their genetic makeup. Finally, he expands the view to encompass factors larger than that one individual. How culture has shaped that individual’s group, what ecological factors helped shape that culture, and on and on, back to evolutionary factors thousands and even millions of years old.”
• Most of these models are “Google-able” but some such as “myths and stories” won’t be immediately understood. In that particular case I am referring to Yuval Harari’s Sapiens book.
• In some cases you will see models appear in two categories, that is intentional and helps provide additional context on its source and development
• I will prove to be woefully inadequate in describing quantum physics (and most of the physical sciences), but I do think there are broad principles contained within that field that are useful in day to day decision making. In those cases, I will be exploring less of the technical side and more of the takeaways.

While not explicitly stated, the following writing by George Soros summarizes what I believe are the need and what of mental models. “The complexity of the world in which we live exceeds our capacity to comprehend it. Confronted by a reality of extreme complexity, we are obliged to resort to various methods of simplification: generalizations, dichotomies, metaphors, decision rules, and moral precepts, just to mention a few.” (1)  Put simply, the complexity of reality exceeds our capacity to comprehend it, thus we use mental models to simplify reality. Thus, by definition, a mental model is something that allows us to simplify reality. To be useful, however, not only do mental models have to be simplifications of reality, but they have to closely map to reality….but how close?

A mental model that says the world is flat meets the above criterion of a mental model, but, as we now know, it doesn’t map to reality. Of course, saying the Earth is round also doesn’t perfectly map to reality; however, it is more accurate. Achieving 100% accuracy of reality is impossible and achieving 99.999% accuracy is, as of yet, only possible in the physical sciences (more on this later). This brings to mind an interesting question: to what degree of accuracy is required for a mental model to be useful? In determining what level of accuracy I need, I think through the following: the downside of my mental model being wrong and the upside of my mental model being right. To the merchant, who only has aspirations in his small town, the upside is little and the downside is little of having a mental model that the world is flat. To the scientist who has staked his career on having an opinion on this matter, the upside and downside is much more, thus it would make sense for him to figure out a way to get a more accurate representation.

Let me pause and summarize: The world is too complex to fully comprehend. Mental models allow us to simplify reality and, therefore, interact with reality. No mental model is 100.000% accurate. Accuracy of a mental model falls on a probability scale of 0.001% to 99.999% (throw in an infinite amount of zeros and an infinite amount of nines in there, respectively). The level of accuracy required for a mental model to be useful is dependent on the downside and upside associated with using that mental model. That upside and downside will be unique to every individual, thus the level of accuracy needed will be unique to every individual.

If the level of necessary accuracy of every mental model is related to each individual’s personal upside / downside, then is explaining any mental model useful? Yes, I believe it is still helpful, caveated with as long as the explained mental model points out where the fog remains. Given that mumbo jumbo, (I had to read it a few times to make sure I even followed the logic…certainly point out if/where it is wrong) my goal will to be explain my mental models and (to the extent I know my knowledge gaps…you don’t know what you don’t know) point out where my fogginess remains. It is then up to you to determine if incorporating the mental model as-is is sufficient for your upside/downside potential or if it requires further digging.

A couple caveats: You don’t know how accurate your model is. You don’t perfectly know your potential upside and downside. You don’t know what you don’t know. Protect against this by being intellectually flexible (it’s OK to be wrong – reward yourself when you admit being wrong) and intellectually curios.  

(1) https://www.georgesoros.com/essays/fallibility-reflexivity-and-the-human-uncertainty-principle-2/