The Pit of Expertise
Jan 2024 - Alex Alejandre

Recently on /slatestarcodex a poster described his feelings of being unmoored. Sometimes, we don’t have the answer. The greater our understanding of a field, its debates and arguments, the less we can confidently believe one wholeheartedly. Each argument is well met by a tried and true counterpoints and we start to understand Socrates' Ignore academic questions of attribution “I know I know nothing.”

This video illustrates it in 45 seconds.

We have lost perspective overoptimizing. If only we could start with a goal and stop when we reach it… Stepping back from the bark, we see a tree. Stepping further back from the trees, we remember we’re in a forest. We can answer many questions. We just forget people ask those questions, because we overestimate what everyone knows about our field

Going deeper, when distressed we seek answers to bring us peace, but there’s nothing there. For example in wars, there are moral dilemmas and many horrors. You can lie to yourself, accept the darkness in human societies or resort to other copying mechanisms, only rarely can you find certainty in truth (and other sides obviously disagree.) (N.b. one failure mode is how many use worry as a tool, trying to predict the future and feel control. It never works.)

How can we manage uncertainty then? Symbolic AI floundered here. We can give them probabilities, confidence scores, severity scores (impacts of something being wrong (will it kill this tree, or the whole forest?) akin to risk in value investing). Rationalists flaunting Bayesian reasoning will even just guess numbers, because the proper conditionality framework (what depends on what) yields more explanatory power. Modeling this as a normal distribution Aufhebung, we can build many-dimensional feasibility fields, overlapping to show us the realm of future possibility. Fate holds us in these rails. We choose what to do with it. Quoting myself:

To spy all time and space, existence too,
As if t’were but a mountain range in view.


Addendum

Some posit rational discourse doesn’t work (seen in glorious flame wars before 2016.) Gödel’s first incompleteness theorem shows no system can prove itself, so relying on evidence alone fails to establish consensus. There are outside truths. Approaching them, we’re stuck in logical selfreference. The old hope was that open debate, freed of gatekeepers (e.g. newspaper editors, royal censors), deplatforming, economic costs of correspondence (printing pamphlets, newspapers etc.) etc. would lead to the truth and a new renaissance. Instead, once convinced, politicking, building coalitions and institutions etc. let you force your beliefs on society. They discovered politics:

All human endeavors form a complex emergent phenomena where every action is a consensus decision, arrived at by many actors for contradictory reasons(nothing has ever happened for but one reason.) We all have different hopes and goals after all. Zero sum power arises from the control asymmetry between agents over relevant outcomes, which they barter into a cohesive society-wide movement. A ship with a thousand propellers in every direction still moves somewhere.

When someone relegates belief building to another, he surrenders agency. hitching his wagon to a movement, religion, ideology etc. which saves him from thinking for himself.