On Honesty

In many cultures, honesty is seen as a virtue, an ideal that one should strive for because it holds value in and of itself. However, in many other cultures, including one that I grew up in, bending the truth and information asymmetries are an essential aspect of interacting with others. In some cultures, it may be rude to point out truths that may upset someone, and we routinely have to filter and transform ideas as we share them with the world. Similarly, in some cultures, exaggerations, lies and altered views of reality are so prosaic that they become part of the everyday life and how we interact with others.

Over the course of this essay, I’m going to motivate why I think honesty is an extremely hard principle to understand and adhere to, and what I view as an optimal policy for executing on a reasonable interpretation of honesty.

The Misconception

The conventional view of honesty is that there are facts that are known by you, and you choose to relay some version of them to others. The more your relayed message is alike the facts, the more honest of a person you are. Furthermore, because we believe people have a fundamental right to knowing the truth about their environment, the more of an honest person you are, the more you are providing that facility to your surroundings and thus, the better of a person you are.

Due to many factors and forces, throughout youth, I did not internalize this perspective of honesty and so I never learned to view it as an end in and of itself because it was inherently incompatible with my model of human interaction. In my school of thinking, there are no absolute truths and lies; there are simply various ways of capturing and interpreting perceived events. There is no objective truth or fact; there are simply 1) events that take place, 2) people who perceive them, and 3) personalities that interpret their meaning.

In this world, honesty is the degree to which you communicate out portions of each of the three. However, it’s hard to believe that fully communicating all three levels is necessarily a reasonable policy, because the way you perceive and interpret events is shaped by a set of idiosyncratic biases and priors that shape your world-view, and there is no reason why they should be useful to someone else. So we must hold back some of our thoughts about an event; but which ones? A reasonable stance is that we should always be honest about 1 and more conservative with 2 and 3 since 1 is a more objective capturing of external facts about the world, free of our biases. While this is reasonable, it assumes that we have a good sense of the divide between 1 and 2, which is not always the case: the way we remember events is greatly influenced by how we perceived and interpreted them, so separating 1 from 2 and 3 is actually not a trivial task.

This suggests that one cannot have a simple “always be honest” policy. How you present information of type 1, 2 and 3 externally needs to go through scrutiny at runtime and be subjected to some evaluation function (for ex. maximizing expected happiness for yourself and your environment over a certain horizon).

There are times when you have learned misinformation about an event, or have reasoned incorrectly about a premise, or have been misled by someone (intentionally or accidentally) to think about an issue through an incorrect lens (for some definition of correctness). Sometimes, especially in professional settings, you are operating under uncertain conditions and it is important that you do not reveal all the information you possess to everyone, or else you risk propagating uncertainty and instability throughout the organization. In many of these situations, we are taught to withhold information and be intentional about what information we project to our environment.

However, as I’ve gotten older, I’ve started to realize the tremendous value of completely short-circuiting this entire exercise and simply having the goal of communicating out the most clear picture of your internal representation and memory of the world, and letting others do the work of ignoring what they don’t like and discounting what they think is unreasonable. To motivate why I think this is an optimal policy, let’s take a quick detour.

A Detour on Quantum Field Theory

There are many incredibly intelligent physicists and cosmologists actively researching theories that help us describe our world at the quantum level and what aspects of it we should think of as being fundamental vs. emergent. In a paper published in 2019, a group of theoretical physicists at Stanford and Caltech provide a theory that aims to describe the cosmological evolution in terms of relatively simple quantum circuitry operating on quantum information.

While I don’t have a good understanding of much of the technical discussion, the intuition provided in the paper and related presentations is one that I find deeply satisfying about the nature of the universe. I have no reason to believe that this formulation (and associated assumptions) is one that correctly captures fundamental aspects of the universe at a quantum level, but nevertheless, I find its approach useful for reasoning about more practical scenarios in my own life.

This intuition is fairly simple. It starts by associating a finite number of degrees of freedom with any region in space. This number defines the total number of ways that the quantum particles and fields in that region of space can arrange themselves; a larger region of space necessarily has more degrees of freedom. It then asserts that if we believe the observation that space is expanding, it means in totality we are increasing the number of degrees of freedom associated with space-time. So either the universe is not a closed system (the closed system would be the universe + whatever object these degrees of freedom are coming from) or that not all degrees of freedom in the physical universe are associated with space-time is simply a part of the degrees of freedom in the universe. The second explanation is what the paper argues. Specifically, it maintains that at the beginning of the age of the universe (Big Bang) the almost all qubits were in their zero-state and all quantum fields untangled. As time goes on, more and more of these quantum objects are interacting with circuit of the universe and becoming entangled, and becoming part of the fabric of space-time. In the limit of time, all of these quantum objects will be entangled with the circuitry of the universe and there is no more information that can be captured, signaling the death of the universe.

An essential point in this view is the central role that the transfer of quantum information plays in the evolution of the system. In this theory, “empty space” is actually occupied by large numbers of quantum particles in their zero state, and with passage of time (not really time, but close enough..) information is transmitted between neighboring particles, increasing their degree of mutual information, a measure used to quantify entanglement, which in turn, grows the fabric of space.

Back to Honesty

So how could this possibly be related to honesty? The way that I have started to think about honesty is that you can model people’s perception of reality as neighborhoods in this space. As time goes on, more information is shared between different neighborhoods, and more conclusions are reached because of information found in the union of multiple neighborhoods, causing these neighborhoods to both grow and become more and more entangled with one another.

Suppose there’s a single event \(e\), observed by \(O_1\). When \(O_1\) comes across another agent \(O_2\), she has the choice to share any information (of type 1, 2 and 3 from above) about \(e\) with \(O_2\). Using the model from the previous section, we can think of \(O_1\) sharing information with \(O_2\) about \(e\) as entangling certain part of \(O_2\)’s circuitry with that of \(O_1\) because now \(O_2\)’s model of the world shares information with that of \(O_1\).

It’s easy to see if \(O_1\) chooses to disclose information that is directly inconsistent with fundamental characteristics of \(e\), she is explicitly causing \(O_2\)’s understanding of \(e\) to be biased and inaccurate. I argue that even witholding information about \(e\) (i.e. describing \(e'\), $some proper subset of her knowledge about \(e\)) is problematic, because while the information withheld is additive to \(e'\), there is no guarantee that it will be additive to whatever \(O_2\) interprets \(e'\) as. In the process of interpretation of the event, she may make assumptions and inferences that are directly in conflict with \(e - e'\), which means by choosing to communicate a proper subset of the event, \(O_1\) ended up communicating the description of a different event.

Whether through ommission or lies, by communicating information to \(O_2\) about \(e\) using a different description of \(e\), \(O_1\) is betting that \(O_2\) will not share information with any other observer of \(e\) in any subsequent interaction.

When this happens, \(O_1\) is effectively betting that \(O_2\) will not be entangled with any other observer of \(e\) in any future timestep of existence. Each bit of mischaracterization that \(O_1\) projects into the world makes certain configuration of the future states of the universe undesirable ones (specifically those in which \(O_2\) will learn mutually inconsistent information). The more mischaracterizations, the more entanglements become undesirable, which makes the set of possible good future states more and more limited.

However, \(O_1\) has very little control over the evolution of the universe and so the future unflods as it does, future entanglements happen as they do, leaving subjects that have restricted more and more entanglements in worry, pain and anxiety.